Category: Gold / XAU-USD Basics

There is a particular kind of expensive confusion that happens on inflation day. The Consumer Price Index comes in hot, the headlines say inflation is back, gold does something for ninety seconds, and then it does the opposite. You conclude that the market is irrational. It is not. Part of what happened is that you and the Federal Reserve were looking at two different inflation numbers. So, what is PCE in gold trading, and why does the measure nobody quotes matter more than the one everybody quotes? The short answer is that the Fed’s two per cent target is not written in terms of CPI at all. It is written in terms of PCE, and the two disagree far more often than most traders realise.

This article does the arithmetic. I pulled the monthly Personal Consumption Expenditures price index from the Bureau of Economic Analysis, 812 months from January 1959 to August 2026, and the monthly Consumer Price Index from the Bureau of Labor Statistics, measured the gap between them month by month, then checked which of the two actually lines up with how gold moves. The results are not close.

What is PCE in gold trading, in the only sense that matters

PCE stands for Personal Consumption Expenditures. The price index built from it is produced monthly by the Bureau of Economic Analysis, and it measures what Americans actually spent on goods and services, including spending made on their behalf by employers and government programmes. CPI, by contrast, measures the cost of a fixed basket bought directly out of pocket by urban consumers.

That difference in construction sounds like a statistician’s quibble. It is not, because of one sentence. In its Statement on Longer-Run Goals and Monetary Policy Strategy, the Federal Open Market Committee writes that it “reaffirms its judgment that inflation at the rate of 2 percent, as measured by the annual change in the price index for personal consumption expenditures, is most consistent over the longer run with the Federal Reserve’s statutory maximum employment and price stability mandates.” You can read that statement in full on the Federal Reserve’s own site.

Read it slowly. The target is two per cent. The yardstick is the PCE price index. CPI is not mentioned. So when a CPI print lands and you reason forward to what the Fed will do, you are using a proxy for the committee’s actual measure without knowing how big the error on that proxy is.

That is why this matters to a gold trader. Gold pays no coupon and no dividend, so what it competes against is the real return available on cash and bonds, roughly the nominal rate minus inflation. The chain runs from inflation data, to what the market expects the Fed to do, to real rates, to the opportunity cost of holding a metal that yields nothing. If the first link is a measure the Fed does not target, the chain is weaker than you think.

How far apart are the two numbers, actually

Here is the measurement. Take the twelve month inflation rate on both measures, the figure quoted in every headline, and compare them in the same month. Across 703 months from January 1968 to August 2026:

• CPI inflation runs above PCE inflation by a mean of 0.508 percentage points and a median of 0.437 points.
• CPI is the higher of the two in 83.9 per cent of all months.
• The two measures differ by at least 0.3 points in 66.0 per cent of months, by at least 0.5 points in 47.4 per cent of months, and by at least a full percentage point in 15.2 per cent of months.
• The widest the gap has ever been is 3.74 points, in June 1980. The most negative was minus 1.78 points, in August 1983.

Sit with the third line. In roughly one month out of every two, the two official inflation measures for the same economy in the same month differ by half a percentage point or more. These are not rounding errors. Half a point is the difference between a number that sounds tolerable and a number that sounds like a problem.

The bias is persistent and it has a known cause. CPI holds its basket relatively fixed, while PCE reweights as spending patterns shift, so when people substitute away from whatever is getting expensive, PCE captures that and CPI largely does not. PCE also has broader scope, most notably in health care, where it counts spending made on a consumer’s behalf. The Bureau of Economic Analysis sets out the construction on its PCE price index page. The direction of the bias is not random, which is why CPI sits on top more than four times out of five.

The same month, read two ways

Abstractions are easy to nod along to, so here are the last few months of data as they actually arrived. All figures are twelve month rates, in per cent.

• April 2026: CPI 3.95, PCE 3.57, core PCE 3.08. Gap of 0.37 points.
• May 2026: CPI 4.27, PCE 3.82, core PCE 3.16. Gap of 0.45 points.
• June 2026: CPI 3.73, PCE 3.45, core PCE 3.04. Gap of 0.27 points.
• July 2026: CPI 3.54, PCE 3.36, core PCE 2.98. Gap of 0.18 points.
• August 2026: CPI 3.71, PCE 3.42, core PCE 3.01. Gap of 0.29 points.

Look at May. A trader watching CPI saw 4.27 per cent, a number comfortably more than double the target, the sort of print that gets described as inflation running out of control. The committee’s own measure said 3.82 per cent, and the core version of it, which strips food and energy, said 3.16 per cent. Those are three materially different stories about the same month, and the one that governs policy is the one that was quoted least. Note too how much steadier core is across those five months: it moves between 2.98 and 3.16 while headline CPI swings from 3.54 to 4.27.

Why the Fed watches the boring version

Across all 800 months for which both are available, the standard deviation of headline PCE inflation is 2.382 percentage points, against 2.084 points for core PCE, which excludes food and energy. The sharper difference is month to month: the median change in the headline twelve month rate from one release to the next is 0.149 percentage points, against 0.086 points for core. The headline figure jumps around roughly 1.7 times as much between releases as the core figure does.

That is the practical case for core. It is not that food and energy do not matter to households, because obviously they do. It is that they swing for reasons monetary policy cannot address within the horizon over which policy works, so a committee trying to read the underlying trend gets a cleaner signal from the series without them. When officials talk about the trend, they are usually talking about core PCE.

One more figure worth keeping in perspective. Of those 800 months, the PCE twelve month rate sat within a quarter of a point of the two per cent target in only 12.5 per cent of them, and in 19.7 per cent of months since 1995. For about four months in five, inflation is not at target. Being away from target is the normal state of the world, not a crisis.

What gold actually tracks, measured

Now the part that connects this to a chart. I took monthly gold returns from the World Bank’s commodity price series, the monthly ten year US Treasury yield from the Federal Reserve’s H.15 release of selected interest rates, and PCE inflation from the BEA, over the 283 months from February 2003 to August 2026. Then I built a crude real yield as the nominal ten year yield minus the trailing twelve month PCE inflation rate, and measured how monthly gold returns correlate with changes in each quantity.

• Change in the real yield: correlation minus 0.308, which is an R squared of 9.5 per cent.
• Change in the nominal ten year yield: correlation minus 0.263, an R squared of 6.9 per cent.
• Change in PCE inflation on its own: correlation plus 0.152, an R squared of 2.3 per cent.

The ordering is the finding. Gold’s monthly behaviour lines up with the direction of real yields about four times more strongly, measured by variance explained, than with the direction of inflation by itself. The sign on the real yield is negative, which is what the opportunity cost argument predicts: when the real return on bonds rises, the case for holding a metal that pays nothing weakens.

This is exactly why a hot inflation print does not mechanically send gold up. A hot print raises inflation, which taken alone is the plus 0.152 column and faintly supportive. But the same print can raise the expected path of policy rates by more, which lifts the real yield, and that is the minus 0.308 column and the stronger effect. Two forces, opposite directions, and the stronger one often wins. That is a better explanation for the ninety second move that reverses than irrationality.

What this does not say

An R squared of 9.5 per cent is a real relationship and a weak one. Roughly ninety per cent of gold’s monthly variation comes from things not in this calculation, so anyone telling you real yields determine the gold price is overstating what I can measure. The honest claim is narrower: of the three quantities tested, real yields are the one gold attends to most, and inflation alone the one it attends to least.

My real yield is also computed after the fact, from inflation already published. The real yield the market actually prices is forward looking and best read from inflation protected securities, which is a different series. I used the cruder version because it can be built entirely from two primary sources a reader can check, and I would rather flag the approximation than hide it. Monthly data cannot see the first minutes after a release either, so nothing here measures an intraday reaction.

Finally, none of this is directional. There is no state of the inflation data that tells you what gold will do next. The arithmetic explains why the relationship you were taught is unreliable, and it does not replace it with a better rule, because there is not one, and no entry, stop or target discussed should be treated as a signal.

How I would actually use this

Three things follow, and none of them is a trade.

First, when you are reasoning about what the Fed might do, reason about PCE, and preferably core PCE, because that is the measure in the committee’s own mandate sentence. CPI is a decent proxy with a known bias of about half a point and an error that exceeds a full point in one month out of seven. Use it knowing that.

Second, stop expecting inflation prints to push gold in a fixed direction. Gold attends to real yields roughly four times more than to inflation on its own, and a single print moves both at once. Expecting a clean reaction to a messy, two sided input is how people end up convinced the market is rigged against them.

Third, treat PCE day as a scheduled event in the same way as CPI day and FOMC day. It arrives monthly, the date is published in advance, and the only cost you can genuinely control is the cost of being surprised by a calendar you could have read a week earlier.

So that is the answer to what is PCE in gold trading. It is the measure that actually defines the Fed’s two per cent target, it runs on average about half a point below the CPI figure in the headlines, it disagrees with that figure by more than a full point in fifteen per cent of months, and its core version is the steadier series officials lean on. If you trade around inflation data, watch the number the decision makers are watching.

Frequently asked questions

What is PCE in gold trading, in one sentence?

It is the Personal Consumption Expenditures price index, the inflation measure the Federal Reserve actually uses to define its two per cent target, which makes it the more relevant of the two official inflation numbers when you are reasoning about interest rates and therefore about the opportunity cost of holding gold.

Is PCE more important than CPI?

For working out what the Fed is likely to do, yes, because the committee’s own mandate sentence names PCE and does not name CPI. For understanding what households are paying out of pocket, CPI is the closer fit. They answer different questions, which is why both are published.

How different are CPI and PCE in practice?

Over 703 months to August 2026, CPI inflation ran above PCE inflation by a mean of 0.508 percentage points, and was the higher number in 83.9 per cent of months. The two differed by at least half a point in 47.4 per cent of months and by more than a full point in 15.2 per cent of months.

Why did gold fall when the inflation number came in hot?

Most likely because the print moved two things at once. It raised inflation, which on its own is mildly supportive of gold, but it also raised the expected path of interest rates, which lifts real yields, and the measurements in this article show gold is roughly four times more responsive to real yield changes than to inflation changes alone. The stronger force often wins.

What is core PCE and why do officials keep mentioning it?

Core PCE is the same index with food and energy excluded. It is the steadier series: its twelve month rate moves by a median of 0.086 percentage points between releases against 0.149 points for the headline measure, so it gives a cleaner read on the underlying trend that policy can actually influence.

What Gold Empire actually does

Gold Empire is a free channel about surviving in gold rather than getting rich in it. There is no signal service here and nothing is sold on the promise of a return. The free Gold Survival Sheet is a single page on protecting an account through exactly the kind of week this article describes, and the only thing it asks for is an email address you can unsubscribe from. The risk management guide that sits underneath all of this is simply published, as is the rest of the scheduled macro calendar series: what CPI is in gold trading, what the FOMC is and why gold reacts, and what nonfarm payrolls means in gold trading.

The model is deliberate. Follow along for free, read the arithmetic, and decide for yourself. If a piece of writing ever makes you feel you must act immediately, that is a reason to be suspicious of the writing, including mine.

About the author

Matthew writes Gold Empire. He spent years assuming the inflation number on the news was the inflation number that mattered, which is a cheap mistake to fix and an expensive one to keep, and he now reads the committee’s own wording instead of the headline. He has no interest in telling anyone where price is going. You can read more about the approach on the about page.

---

This article is educational and is not financial advice. Trading gold carries a high risk of losing money rapidly, and no entry, stop or target discussed should be treated as a signal. Figures are computed from the published Personal Consumption Expenditures price index from the Bureau of Economic Analysis, the Consumer Price Index from the Bureau of Labor Statistics, the Federal Reserve H.15 interest rate release and the World Bank monthly commodity price series, as described above, and were accurate at the time of writing. Past behaviour of any data series is not a prediction of its future behaviour. Decide your own risk, and never trade money you cannot afford to lose.

Four names, two ideas, and one of the four can cost you more than you agreed to lose. Most people learn them in the wrong order, which is to say they learn the two buy orders when they are opening a position and never think about the two sell orders until one of them fills somewhere they did not expect. If you have searched what is buy limit sell limit buy stop sell stop, you are probably staring at a platform dropdown with four options and no explanation of why they are not interchangeable.

They are not interchangeable. Two of them promise you a price and may never fill. Two of them promise you a fill and make no promise at all about the price. That single distinction is the whole subject, and the reason it matters is that your protective stop on a gold position belongs to the second group.

Everything numerical below is either quoted from the European regulator I link, or arithmetic I work out in front of you from assumptions I state. There are no gold prices here, nothing is a forecast, and nothing is a suggestion about what you should place or where.

What is buy limit sell limit buy stop sell stop, in one sentence each

Strip the jargon and every pending order is the same instruction with two variables: which side you want to end up on, and whether you want the market to come to your price or to prove something first.

A limit order sits on the side of the market that is better for you than the current price. You are waiting for a move in your favour before you transact. A stop order sits on the side that is worse for you than the current price. You are waiting for the market to move against your current position, or to break out in a direction you want to join, before you transact.

• Buy limit. Placed below the current price. You will buy, but only if price falls to your level. You are asking for a discount.
• Sell limit. Placed above the current price. You will sell, but only if price rises to your level. You are asking for a premium.
• Buy stop. Placed above the current price. You will buy once price trades up through your level. You are asking for confirmation and paying for it.
• Sell stop. Placed below the current price. You will sell once price trades down through your level. This is where almost every protective stop on a long position lives.

I wrote at length about the two buy orders and what each one costs in a fast market in buy stop and buy limit explained. This article is the complete set, and the half that gets skipped is the sell side, because that is the half holding your risk.

The promise each order actually makes

Here is the part that platform documentation states in a footnote and that costs people real money.

A limit order has a price guarantee and no fill guarantee. If your buy limit sits below the market and price never reaches it, nothing happens. If price reaches it but there is not enough size at your level, you may get part of your order or none of it. What cannot happen is a fill at a worse price than you specified. The limit is a floor under the quality of your execution. I set out what the unfilled case actually looks like in what happens if a limit order is not filled.

A stop order is the mirror image, and the mirror is unkind. A stop has a fill guarantee and no price guarantee. Once price trades through your stop level, the order converts into a market order and takes whatever the book offers next. In a quiet hour that is a few cents away from your level. In the first second after a scheduled release, or at the Sunday open after a weekend of news, the next available price can be a long way from where you asked.

That is not a platform defect and it is not your broker being difficult. It is the definition of the order. A stop that refused to fill at a worse price would not be a stop, it would be a stop limit, and a stop limit that goes unfilled during a fast move leaves you in the position you were trying to exit. There is no version of this where you get both guarantees.

What the overshoot costs, in units you can plan with

Dollars per ounce are useless for this because they depend on your size. So measure the damage the way a risk plan measures everything, in multiples of the amount you decided to risk on the trade. Call that amount one R.

Set up the arithmetic. You are long. Your protective sell stop sits a distance below your entry, and that distance is what defines one R. If the market gaps past your stop, your fill lands further away, and your realised loss is the full distance divided by the planned distance. If the gap carries price a quarter of the stop distance past your level, you lose 1.25R. Half the stop distance past, 1.50R. A full stop distance past, 2.00R. Twice the stop distance past, 3.00R.

Read that table once more and notice what it is telling you. A tighter stop does not protect you from a gap, it makes the gap more expensive in R terms, because the same overshoot in dollars is a larger multiple of a smaller planned distance. The size of the gap is set by the market. The size of the denominator is set by you, before you enter. That is the only lever in the equation that belongs to you, and it is the argument underneath everything in the risk management guide.

Where the regulator draws the line, and what that tells you about the tail

Gaps are not hypothetical, and the European Securities and Markets Authority built an entire set of rules around the fact that retail accounts were being carried past zero by them. When ESMA announced its product intervention measures on contracts for difference, it set a leverage cap of 20:1 for gold, alongside 30:1 for major currency pairs and 10:1 for other commodities. It also imposed a margin close out rule standardised at 50 per cent of the minimum required margin, and negative balance protection on a per account basis.

Turn the leverage cap into the number that matters. A 20:1 cap means the minimum required margin on a gold position is 5.00 per cent of its notional value. The close out rule bites when account equity falls to half of that, which is 2.50 per cent of notional. So an account funded with exactly the minimum margin has a total adverse move of 2.50 per cent of position value between fully funded and forced liquidation. Not 5 per cent. Half of it.

Two things follow. The first is that a sell stop placed further than 2.50 per cent of notional away from entry, on an account at minimum margin, will never be the thing that closes your position, because the broker’s close out arrives first. The second, and more important, is why negative balance protection had to be written into law at all. Regulators do not legislate against risks that stay inside the agreed loss. They legislate against the ones that run past it, and a stop order filling past its level is exactly that shape of risk.

Those rules apply to retail accounts under European rules. If your account sits elsewhere, the arithmetic above still holds, but the floor underneath it may not, and that is worth knowing before you need it.

Where each of the four belongs in a survival plan

Now the practical part, written as mechanics rather than instruction.

Buy limit and sell limit are entry tools for patient positions. They are the orders you use when your thesis is that price is currently wrong and will come to you. Their failure mode is the quiet one: you do not get filled, you watch the move go without you, and you feel the itch to chase. Nothing is lost except the opportunity, and a missed entry has never emptied an account.

Buy stop is an entry tool for breakout positions. You are deliberately paying a worse price in exchange for the market demonstrating something first. Its failure mode is slippage on entry plus the risk that the breakout was the high of the move. The cost is real but it is bounded by the stop you attach to it.

Sell stop does two completely different jobs and this is the confusion worth clearing. As an entry, it opens a short on a breakdown. As an exit, it is the protective stop on every long you hold. The order type is identical. The consequence of a bad fill is not. A slipped entry costs you a worse starting price on a trade you chose to take. A slipped exit costs you more than the loss you signed up for on a trade that is already going wrong.

Which is why the planning question is not which order type is best. It is: given that my protective order can fill past its level, is my position size small enough that a 2R or 3R outcome is survivable rather than structural? If the honest answer is no, the position is too large, and no choice of order type fixes that. Gaps around the weekend are the most predictable version of this, and I went through the mechanics in what is a weekend gap in gold trading.

Four checks before you place any of the four

These take under a minute and they are the difference between an order you understand and one you are hoping about.

1. Say the guarantee out loud. Limit means my price or nothing. Stop means a fill at whatever is next. If you cannot say which one you just placed, you do not yet know what you placed.
2. Check the side. Buy limit below, sell limit above, buy stop above, sell stop below. Platforms reject an order placed on the wrong side of price, and traders routinely misread that rejection as a broker problem rather than a sign they wanted the other order.
3. Price the bad fill before you need it. Ask what a 2R outcome on this position does to your account. If the answer is uncomfortable, reduce size now, not after.
4. Look at the calendar. A pending order left running into a scheduled release or over a weekend is a different order than it was when you placed it, because the distribution of next available prices has changed.

What this article is not saying

I want the limits stated plainly. The R multiples above are arithmetic on assumptions I chose to make the mechanism visible. They are not an estimate of how often gold gaps, by how much, or in which direction, and I have deliberately not offered one, because a credible figure would need a dataset and a method, not a confident sentence.

Nothing here tells you which order to use, where to place it, or when. The regulatory figures are quoted from ESMA and apply to retail accounts under European rules at the time the measures were announced. Rules change, brokers differ, and your account may sit under a completely different regime.

Frequently asked questions

What is buy limit sell limit buy stop sell stop, in the shortest possible answer?

Limits sit on the favourable side of price and guarantee your price but not a fill. Stops sit on the unfavourable side and guarantee a fill but not your price. Buy limit is below, sell limit is above, buy stop is above, sell stop is below.

Which of the four can lose me more than I planned?

The two stop orders, because they become market orders when triggered. A sell stop used as a protective exit is the one that matters most, since a fill past the level turns a planned 1R loss into 1.25R, 1.50R, 2.00R or more depending on how far price travels before the fill.

Is a sell limit the same as a take profit?

On a long position, mechanically yes. A take profit above the market on a long is a sell limit, and it inherits the limit guarantee: you get your price or you do not transact. That is why a take profit never slips in your favour and never slips against you.

Why did my platform reject my pending order?

Usually because the level is on the wrong side of the current price for the type you chose, or because it is inside the minimum distance your broker enforces around the market. Both are rules of the order type and the venue rather than faults.

Does a stop limit solve the slippage problem?

It replaces one problem with another. A stop limit will not fill worse than your limit price, but if the market moves straight past that limit during a fast move, it does not fill at all, and you keep the position you were trying to exit. You are choosing which failure you would rather have.

How much margin does a gold position need?

Under the ESMA measures the retail cap for gold is 20:1, which puts minimum required margin at 5.00 per cent of notional, with the close out rule triggering at half that, 2.50 per cent. Outside that regime the numbers are set by your broker and can differ substantially.

What Gold Empire actually does

Gold Empire is a free channel about surviving in gold rather than getting rich in it. There is no signal service here and nothing is sold on the promise of a return. The free Gold Survival Sheet is a single page on protecting an account through exactly the kind of fast market this article describes, and the only thing it asks for is an email address you can unsubscribe from. Everything else, including the risk management guide underneath all of it, is simply published.

The model is deliberate. Follow along for free, read the arithmetic, and decide for yourself. If a piece of writing ever makes you feel you must act immediately, that is a reason to be suspicious of the writing, including mine.

About the author

Matthew writes Gold Empire. He learned the difference between the price guarantee and the fill guarantee the expensive way, on a Sunday open, and has written the arithmetic down ever since rather than relearning it. He has no interest in telling anyone where price is going. You can read more about the approach on the about page.

---

This article is educational and is not financial advice. Trading gold carries a high risk of losing money rapidly, and no entry, stop or target discussed should be treated as a signal. The regulatory figures are quoted from the European Securities and Markets Authority announcement linked above; the R multiples are arithmetic from stated assumptions, not estimates of real world frequency. Decide your own risk, and never trade money you cannot afford to lose.

If you have been trading gold for more than a month, you have met the eight thirty problem. The chart is quiet, the spread is normal, and then at a fixed minute everything widens at once and your position is suddenly worth something different. Ask what is CPI in gold trading and most answers stop at the definition: it is the Consumer Price Index, a measure of inflation, published monthly by the Bureau of Labor Statistics. True, and almost useless. The definition tells you nothing about the part that actually empties accounts, which is how much a single print moves the number everyone is watching, and how little that one print tells you about the trend it is supposed to measure.

So this article does the arithmetic instead. I pulled the full monthly Consumer Price Index series, 955 observations running from January 1947 to August 2026, and measured the thing nobody measures: the size of the jump in the headline rate from one release to the next, and the gap between what a single month says and what the year actually says. The answers are not subtle, and they change how you should size risk around the calendar.

What is CPI in gold trading, in the only sense that matters

The Consumer Price Index tracks what a fixed basket of goods and services costs over time. The Bureau of Labor Statistics publishes it monthly, on a schedule set a year in advance, at a fixed time in the morning US Eastern time. Two versions matter to anyone watching a screen. The headline number includes everything. The core number strips out food and energy, on the argument that those two swing for reasons that have little to do with underlying inflation.

Gold cares because gold has no yield. It pays no coupon and no dividend, so what it competes against is the real return on cash and bonds, which is the nominal rate minus inflation. When inflation data shifts the market’s view of what the Federal Reserve will do with rates, it shifts the opportunity cost of holding a metal that pays nothing. That is the entire transmission mechanism, and it runs through expectations rather than through the index level itself.

Which is the first thing to understand. The CPI level is not news. It is published, it is known, and by the time you read it the market has already seen it. What moves price is the distance between the number and what the market had already priced. That distance is unknowable in advance, which is why the honest version of this article cannot tell you what to do at eight thirty, only what the distribution of outcomes looks like.

How much does one print actually move the headline rate

Here is the measurement that is missing from almost every explanation of this subject. Take the twelve month inflation rate, the figure quoted in every headline, and ask how much it changes from one monthly release to the next. Across the period since January 2000, which is 318 month to month comparisons:

• The median change in the twelve month rate is 0.238 percentage points.
• The mean change is 0.322 percentage points.
• The largest single month change is 2.631 percentage points.
• The rate moves by at least 0.1 points 79.9 per cent of the time, at least 0.2 points 58.5 per cent of the time, at least 0.3 points 40.9 per cent of the time, and at least 0.5 points 17.0 per cent of the time.

Read that last line again. Four times out of five, the headline inflation rate is a different number than it was a month ago. The thing that is reported as though it were a fixed fact about the economy is in practice a quantity that wobbles by a quarter of a point most months, and occasionally by much more. The market knows this, which is why the release is scheduled as a risk event rather than treated as a routine update.

One month is not the trend, and the gap is bigger than you think

Now the second measurement, and the more useful one. Take a single month’s change in the index and annualise it, which is what a commentator is doing when they say inflation is running at such and such a pace. Then compare it with the trailing twelve month rate over the same period. Since 2000, across 319 observations:

• The median gap between the one month annualised figure and the twelve month rate is 1.792 percentage points.
• The gap is at least 1 point 71.8 per cent of the time.
• The gap is at least 2 points 43.9 per cent of the time.
• The gap is at least 3 points 28.8 per cent of the time.

In other words, the story a single print tells disagrees with the story the year tells by more than a full percentage point in roughly seven months out of ten. The monthly standard deviation of the index change since 2000 is 0.3036 per cent, which annualises to 3.643 percentage points of noise around any single observation. One print is a noisy estimate of a slow variable, and the market reacts to it anyway, because the market is pricing the change in expectations rather than the truth about inflation.

That distinction is the whole reason the first candle after a release so often reverses. The initial move prices the surprise. The minutes after it price the reconsideration of what the surprise actually means. Neither of those is a forecast you can make in advance, and no entry, stop or target discussed should be treated as a signal.

Headline against core, and why both get quoted

The headline and core measures disagree more than most people expect. Comparing the twelve month rates since 2000, the median absolute gap between them is 0.734 percentage points, and the headline figure sits above core in 56.1 per cent of months. As of the August 2026 data, the headline twelve month rate was 3.71 per cent and the core rate 2.76 per cent, a gap of about 0.95 points.

This matters operationally rather than philosophically. Two numbers arrive in the same instant, they frequently point in different directions, and different parts of the market weight them differently. That is part of why the first reaction to a release is often disorderly, and why the spread during that window is the widest it will be all day. You are not being singled out. The liquidity is genuinely thinner, because the people providing it do not know the fair price either until the dust settles.

The calendar is the part you can actually control

Here is the practical turn. You cannot forecast the print. You can know exactly when it lands, and that is a real edge over the version of yourself who gets surprised.

CPI is released twelve times a year. The Federal Reserve’s Open Market Committee met eight times in 2026, on 27 to 28 January, 17 to 18 March, 28 to 29 April, 16 to 17 June, 28 to 29 July, 15 to 16 September, 27 to 28 October and 8 to 9 December, with four of those meetings accompanied by the Summary of Economic Projections, according to the Federal Reserve’s published FOMC calendar. Add roughly twelve employment reports and you are looking at something in the order of thirty two scheduled high attention events in a year. That is about one every eleven days.

Thirty two days out of roughly two hundred and fifty trading days is not a small share of the year, and it is entirely knowable in advance. The survival question is not how to trade those days. It is whether you have decided, before the week starts, what your exposure looks like when one arrives. A position sized for a quiet Tuesday is a different position on a release morning, even though nothing about your analysis changed.

If you want the mechanics of the release window itself, the behaviour of spread and fills during high impact news is covered in more detail in how to trade gold through high impact news. The two other scheduled events in this family have their own pieces: what the FOMC is and why gold reacts, and what nonfarm payrolls means in gold trading. Taken together with this one, that is the full scheduled macro calendar a gold trader actually has to plan around.

What the data does not say

I want to be exact about the limits here, because this is the point where articles like this usually overreach.

Nothing above measures how gold responds to CPI. It measures how CPI itself behaves, which is a different and more reliable thing. I have not computed an average gold move on release days, and I would be sceptical of anyone who quotes you one without showing the method, because the answer depends enormously on the window you choose, the period you sample and how you handle the days when something else was also happening.

Nor does any of this tell you direction. A hot print does not mechanically send gold one way. The same number can be read as inflationary, which historically supports gold, or as something that will force higher rates, which raises the opportunity cost of holding it. Both readings are available in the same instant, which is precisely why the first move reverses so often.

The numbers in this piece come from the monthly Consumer Price Index for All Urban Consumers, seasonally adjusted, produced by the Bureau of Labor Statistics and retrieved from the Federal Reserve Bank of St Louis’s published CPIAUCSL series, together with the core series that excludes food and energy. Every percentage in this article was computed from that data rather than recalled, and the method is stated beside each figure so you can reproduce it.

How I would actually use this

Three things follow, and none of them is a trade.

First, put the release dates in whatever you plan your week in. Not to trade them, but so that no release is ever a surprise. The entire cost of being surprised is paid in the first sixty seconds, and it is avoidable for free.

Second, decide your release day exposure once, in advance, in writing, when nothing is happening. Deciding at eight twenty nine is not deciding, it is reacting, and the arithmetic above says you are reacting to a quantity that disagrees with its own twelve month trend by more than a point most of the time.

Third, stop treating a single print as information about the economy. It is one noisy observation of a slow moving variable, with a standard deviation that annualises to more than three and a half points. The twelve month rate is the signal. The monthly number is mostly the noise around it, and the market trades the noise because the noise is what is new.

That is the honest answer to what is CPI in gold trading. It is a scheduled, measurable, highly repetitive risk event, whose headline figure moves by at least a tenth of a point four times out of five, and whose monthly reading disagrees with its own annual trend by more than a point seven times out of ten. You cannot predict it. You can absolutely plan around it, and the planning is where the survival is.

Frequently asked questions

What is CPI in gold trading, in one sentence?

It is the monthly Consumer Price Index release, a scheduled inflation measurement that moves gold indirectly by changing what the market expects from interest rates, and therefore the opportunity cost of holding an asset that pays no yield.

Does a high CPI number always push gold up?

No, and the assumption that it does has cost people a lot of money. A hot reading can be read as inflation, which has historically supported gold, or as a reason rates stay higher for longer, which works against it. Both interpretations compete in the same minute, which is one reason the first move after a release so frequently reverses.

How much does the inflation rate change between releases?

Since 2000 the twelve month headline rate has changed by a median of 0.238 percentage points from one month to the next, moving at least 0.1 points in 79.9 per cent of months and at least 0.5 points in 17.0 per cent of months. The largest single month change in that period was 2.631 points.

Should I close my positions before CPI?

That is a decision about your own risk tolerance and your own plan, not something an article can answer for you, and nothing here is a recommendation. What the data supports is narrower and more useful: the decision should be made in advance rather than in the minute before the release, because the release window is reliably where the spread is widest and the fills are worst.

Is core CPI more important than headline?

Neither is simply more important. They disagree by a median of 0.734 percentage points, and the headline sits above core in 56.1 per cent of months since 2000. Different participants weight them differently, which is part of why the immediate reaction to a release is often disorderly.

What Gold Empire actually does

Gold Empire is a free channel about surviving in gold rather than getting rich in it. There is no signal service here and nothing is sold on the promise of a return. The free Gold Survival Sheet is a single page on protecting an account through exactly the kind of week this article describes, and the only thing it asks for is an email address you can unsubscribe from. Everything else, including the risk management guide that sits underneath all of this, is simply published.

The model is deliberate. Follow along for free, read the arithmetic, and decide for yourself. If a piece of writing ever makes you feel you must act immediately, that is a reason to be suspicious of the writing, including mine.

About the author

Matthew writes Gold Empire. He spent long enough learning the expensive way that scheduled events are the cheapest risk to manage, because they are the only ones that announce themselves in advance, and he now writes the arithmetic down instead of relearning it. He has no interest in telling anyone where price is going. You can read more about the approach on the about page.

---

This article is educational and is not financial advice. Trading gold carries a high risk of losing money rapidly, and no entry, stop or target discussed should be treated as a signal. Figures are computed from the published Consumer Price Index series as described above and were accurate at the time of writing. Past behaviour of any data series is not a prediction of its future behaviour. Decide your own risk, and never trade money you cannot afford to lose.

A member wrote to me in September with one line: “My brother keeps telling me to put money into gold. Is buying gold worth it, or is he just repeating something he heard?” I liked the question because it was honest about not knowing, and because it is almost impossible to answer in the form it was asked. Worth it compared to what, over what period, and to do what job? Nobody can tell you what gold will be worth next year, and anyone who offers to is selling something.

What I can do is show you who actually buys the metal, in what quantity, and for what stated reason. That is a matter of public record, published quarterly, and it is far more useful than a forecast. So this article is a reading of the demand data, plus my own arithmetic on what holding gold costs you while you wait.

Every calculation below is mine, from assumptions I write out as I go. The tonnage figures come from the World Gold Council’s quarterly report, linked where I use them. There is no gold price anywhere in this article, no target, and no recommendation to buy anything.

Is buying gold worth it? Start by asking who is already buying

In the second quarter of 2026, total gold demand including over the counter activity came to 1,268.9 tonnes, unchanged against the same quarter a year earlier. The first half of the year ran to 2,522 tonnes, about two per cent higher than the year before, at a record total value of 380 billion dollars. Those figures are in the World Gold Council’s Gold Demand Trends report for the second quarter of 2026, published on 30 July 2026.

Underneath that flat headline, the composition is the interesting part. Four groups buy gold, and they do not want the same thing from it.

• Central banks and other institutions took 288.9 tonnes in the quarter.
• Bar and coin buyers, which is private individuals holding metal, took 307.1 tonnes.
• Jewellery consumption came to 278.2 tonnes, the lowest quarterly volume since the pandemic.
• Technology, mostly electronics, used 80.4 tonnes.
• Gold backed exchange traded funds went the other way entirely, shedding 44.8 tonnes.

Run the division and central banks were 22.8 per cent of total demand in the quarter. That is my calculation, 288.9 divided by 1,268.9. Close to a quarter of everything bought was bought by institutions whose job is not to make a return.

The buyer who never explains himself in a quarterly call

I want to sit on the central bank number, because it is the one most retail arguments about gold ignore completely.

Look at the run of quarters. Central banks took 177.9 tonnes in the second quarter of 2025, then 226.3, then 208.2, then a sharp slowdown to 56.5 tonnes in the first quarter of 2026, then 288.9 tonnes in the second. Add the last four of those and you get 779.9 tonnes in a year, an average of 195 tonnes a quarter. That addition is mine, from the same published table.

Central banks are not momentum buyers. They do not have a stop loss, they are not trying to beat an index, and they are not going to sell because the chart looked ugly on a Tuesday. They hold gold because it is a reserve asset that is nobody else’s liability. That last phrase is the whole of it. A government bond is a promise from a government. A bank deposit is a promise from a bank. A bar of gold in a vault is not a promise from anyone, which is exactly why it is dull and exactly why institutions keep it.

You can watch this yourself month by month in the World Gold Council’s monthly central bank statistics, which are built from reported official reserve data rather than from anyone’s opinion.

Two sets of buyers walked in opposite directions last quarter

Here is the detail I find most instructive for a private buyer. In the same three months, central banks bought 288.9 tonnes while exchange traded funds shed 44.8 tonnes. The gap between the two is 333.7 tonnes, my subtraction.

The Council attributes the fund outflows to softer gold prices and to shifting expectations for inflation and interest rates, particularly in North America. That is a perfectly rational response from a fund investor, who is holding gold as a position and will reduce it when the case for the position weakens.

So the two groups are not disagreeing about the facts. They are holding gold for different jobs. One treats it as a trade that can be wrong. The other treats it as a reserve that does not need to be right this year. Before you decide whether buying gold is worth it for you, decide which of those two people you are. The answer changes everything that follows, including how much a bad quarter should bother you.

What it costs to hold, which almost nobody counts

Owning gold is not free, and the cost is quiet enough that people forget to subtract it.

Physical metal costs a dealer margin when you buy, a spread when you sell, and either storage or the risk of keeping it at home. A fund charges an annual management fee. That fee is small enough to ignore in any single year and large enough to matter over the holding period a reserve asset implies.

Let me put numbers on it. Assume a fee charged annually against the value of the holding, no trading, and no tax. The fraction of your stake the fee consumes over a period is one minus the quantity one minus the fee, raised to the number of years. On that basis:

• At 0.25 per cent a year, ten years consumes 2.47 per cent of the stake, and twenty years 4.88 per cent.
• At 0.40 per cent a year, ten years consumes 3.93 per cent, and twenty years 7.70 per cent.
• At 0.75 per cent a year, ten years consumes 7.25 per cent, and twenty years 13.98 per cent.

Those are my calculations, and the assumptions are stated above so you can redo them with your own fee. The point is not that 0.75 per cent is scandalous. The point is that a reserve you intend to hold for twenty years should be bought in the cheapest wrapper you can find, because the fee compounds against you for the entire holding period while the reason you bought it sits unchanged.

If you want the comparison between wrappers done properly, gold ETF versus physical gold returns and whether to buy physical gold or gold stocks cover the two choices either side of this one.

Where supply comes in, and why it moves less than you would think

Total supply in the quarter was also 1,268.9 tonnes, matching demand as it must in this accounting. Mine production contributed 965.6 tonnes, up two per cent on the year. Recycling contributed 326.1 tonnes, down six per cent, as softer prices discouraged people from selling old jewellery.

Notice how small those movements are. Mining gold is slow. A new deposit takes many years to reach production, which means supply cannot respond quickly to a change in demand. That is the structural fact behind most of gold’s behaviour, and it is worth understanding before you form a view on whether owning any is sensible.

Recycling is the faster valve, and it works in the direction you would expect: when the metal is dear, more old jewellery comes back to the refiners, and when it softens, less does. The six per cent decline is that valve closing slightly.

What gold does not do, said plainly

This is the part of the answer my member’s brother had skipped.

Gold produces no income. A bar in a vault pays no dividend, no coupon and no rent, and in twenty years it will still be the same bar. Everything you make or lose is the change in what somebody else will pay for it. That is a real limitation, not a technicality, and it is why gold is a poor substitute for assets that generate cash.

Gold is also not insurance in the sense people mean when they say it. It does not reliably rise when equities fall, in any given week or month, and a buyer who expects that precise behaviour will be disappointed on a regular basis.

And gold is capable of long, boring periods of going nowhere while other things rise. If you cannot sit through that without changing your mind, the demand data in this article will not save you, because the behaviour that erodes accounts is not usually the choice of asset. It is the changing of that choice halfway through under pressure.

Trading gold and owning gold are different activities

I should separate two things that get blurred, because most readers of this site arrive through the trading door.

Owning gold is a decision about what fraction of your savings sits in an asset that is nobody’s liability. It is measured in years and it does not involve leverage.

Trading gold is something else. It is a leveraged position on short term price movement, where the dominant risk is not whether gold is a sensible asset but whether your position is too large for your account. The demand numbers in this article tell you almost nothing about that, and I would not want anyone reading a 288.9 tonne figure as a reason to take a position of any particular size. If trading is what you are weighing, how much money to start trading gold is the more relevant piece, and the sizing arithmetic in it matters more than any demand statistic.

Frequently asked questions

Is buying gold worth it for a small saver?

That depends entirely on the job you want it to do, and I will not answer it as a recommendation. What the data supports is narrower: institutions bought 288.9 tonnes in the second quarter of 2026, close to a quarter of all demand, and they buy it as a reserve rather than for a return. If you want an asset that produces income, gold is the wrong instrument. If you want a small part of your savings in something that is not a promise from an institution, that is the case people are actually making, and the sensible size of it is a question for your own circumstances.

Why did funds sell gold in the same quarter central banks bought it?

Because they hold it for different reasons. Exchange traded funds shed 44.8 tonnes in the quarter, which the World Gold Council attributes to weaker prices and to shifting inflation and interest rate expectations. A fund investor holds gold as a position that can be wrong. A central bank holds it as a reserve that does not need to be right this year.

Does the record demand value mean gold is expensive?

No, and this is a common misreading. The first half of 2026 set a record total value of 380 billion dollars on 2,522 tonnes. Value equals tonnes multiplied by price, so a record value can come from a higher price rather than from more metal changing hands. Volume was up about two per cent. Read tonnes when you want to know about demand and value when you want to know about the size of the market.

How much does holding gold cost each year?

It depends on the wrapper. Physical metal costs a dealer margin at both ends plus storage. A fund charges an annual fee. On my arithmetic, assuming the fee is charged yearly against the holding, a 0.40 per cent fee consumes 3.93 per cent of the stake over ten years and 7.70 per cent over twenty. That is not a reason to avoid funds, it is a reason to compare fees when the intended holding period is long.

Is jewellery a good way to own gold?

Jewellery consumption was 278.2 tonnes in the quarter, the lowest since the pandemic, with spending up 14 per cent as prices rose and volumes fell. As an ornament it is its own reward. As a holding it carries a large making charge you rarely recover on sale, so the gap between the buying price and the selling price is wider than for a bar or a fund.

Should I trade gold instead of owning it?

They are different activities and the second is considerably more dangerous. Trading involves leverage, and with leverage the size of the position decides the outcome far more often than the view does. Nothing in this article is a reason to open a position, and no figure here says anything about where the price goes next.

Where Gold Empire fits

Gold Empire is a free education channel about surviving in gold before trying to grow in it. Everything here is free to read and free to follow on Telegram. The one optional piece is the survival sheet above, a single page on protecting an account. We do not promise profits, we do not publish account results, and we do not believe anyone can tell you what your account or your savings will do. If you want the fuller version of the thinking behind this site, risk management for gold trading is the piece everything else hangs off.

About the author. Matthew writes Gold Empire. He bought his first gold for the wrong reason, a forecast he had read and could not have defended, and he has spent the years since learning to ask what a holding is for before asking what it will be worth. Nothing in this article is financial advice or a recommendation to buy or trade, and no entry, stop or target discussed should be treated as a signal. Trading leveraged gold carries a substantial risk of loss and is not suitable for everyone.

A member wrote to me in the spring with a question I get in some form most months. He had four thousand dollars, he wanted to trade gold, and a video had told him futures were the professional route while CFDs were for beginners. Was that true, and should he move? The honest answer took longer than he expected, because micro gold futures vs XAUUSD is not a question about which instrument is more respectable. It is a question about contract size, and contract size is the thing that decides whether a four thousand dollar account survives its first bad week.

So let me do this the way I would do it on paper. What each instrument actually is, what one move is worth in each, and what account each one quietly requires before the arithmetic stops working against you.

Every calculation below is mine, from assumptions I state as I go. The contract sizes and the market figures come from the regulator’s own report, linked where I use them. There are no gold price levels anywhere in this article, and nothing here is a recommendation to trade.

Micro gold futures vs XAUUSD, stated plainly

Gold futures are exchange traded contracts. The full size gold contract on COMEX covers 100 troy ounces. There is also a micro version covering 10 troy ounces, exactly one tenth. Those two sizes are not marketing copy, they are in the Commitments of Traders report published by the CFTC, which prints “contracts of 100 troy ounces” for gold and “contracts of 10 troy ounces” for micro gold next to the open interest for each.

XAUUSD is something different. It is a contract for difference, an agreement with your broker that settles the change in the gold price. Nothing is delivered and no exchange stands in the middle. The common retail convention is that one standard lot is also 100 ounces, with a minimum size of one hundredth of a lot, which works out at one ounce. I want to be careful here: that convention is set by each broker rather than by an exchange, so it is an assumption in what follows and it is the one thing on this page you should check against your own account before relying on it.

That is the whole structural difference. One is an exchange contract in fixed sizes, the other is a private agreement that can be cut much finer.

The number that decides everything: what one dollar of gold is worth to you

Forget the branding and price a single move. If gold moves one dollar an ounce, the full size futures contract moves one hundred dollars, because it holds a hundred ounces. The micro contract moves ten dollars. One hundredth of a CFD lot, under the convention above, moves one dollar.

Those three numbers are the entire argument. They are not opinions about which venue is more serious, they are multiplication.

Now attach a stop to them. Suppose you decide a trade is wrong if gold moves ten dollars an ounce against you. That is a stop distance, not a prediction, and I am choosing a round number to make the arithmetic legible. At that distance the full size contract puts a thousand dollars at risk, the micro contract a hundred dollars, and the one ounce CFD position ten dollars.

Here is where it stops being abstract. The rule I keep coming back to, and the one behind nearly everything on this site, is that no single trade should cost more than about one per cent of the account. Run that backwards and each instrument tells you the account it expects:

• Full size gold futures, one thousand dollars at risk, implies an account of about one hundred thousand dollars.
• Micro gold futures, one hundred dollars at risk, implies an account of about ten thousand dollars.
• One ounce on a CFD account, ten dollars at risk, implies an account of about one thousand dollars.

That is the answer to my member’s question, and it has nothing to do with which instrument is professional. With four thousand dollars, the smallest available position in full size futures would have been a twenty five per cent risk on a single trade. The micro contract would have been two and a half per cent, still more than I would want. The CFD route was the only one where his smallest possible position was smaller than his risk budget.

The chart is worth a moment because it shows the same point from the market’s side. In the CFTC report for positions as of 22 September 2026, there were 412,800 full size gold contracts open and 98,428 micro contracts. So micro gold is 19.3 per cent of the open contracts, nearly a fifth, which sounds substantial. Convert those contracts into metal and the picture changes: 412,800 contracts of 100 ounces is 41.28 million ounces, while 98,428 contracts of 10 ounces is 984,280 ounces. The micro contract holds 2.3 per cent of the open gold.

A lot of participants, very little metal. That is the signature of a contract built for small accounts, and it is the clearest evidence I know that trading gold in small size is ordinary rather than second class.

What the CFD side gives you, and what it charges for it

If finer sizing were the only consideration, CFDs would win for every small account and the article would end here. It is not, and the differences are worth stating without taking sides.

The first is who your counterparty is. With futures, you face a clearing house through a broker, and the contract specification is set by the exchange and reported to a regulator. With a CFD, you face your broker. That is not automatically sinister, but it means the terms are the broker’s terms, and it means the quality of the firm matters more than it does on an exchange. I have written separately about what happens if your forex broker goes bust, which is the question this leads to.

The second is leverage, and here the regulation is explicit. Under the product intervention measures agreed by ESMA, the European Securities and Markets Authority, retail CFD leverage on gold is capped at 20 to 1, which is the same as saying you must post 5 per cent of the position’s value as margin. The same measures set a margin close out at 50 per cent of required margin and require negative balance protection on a per account basis. Those rules apply to retail clients of EU regulated firms, so whether they cover you depends on where your broker is authorised.

That same ESMA announcement contains the number I would put on the wall of anyone starting out. The national regulators found that between 74 and 89 per cent of retail accounts typically lose money on CFDs, with average losses per client ranging from 1,600 to 29,000 euros. Three quarters to nine tenths. The instrument you pick moves that figure far less than the size you trade and the risk you take per trade.

The third difference is the running cost. A futures position has no overnight financing charge in the way a CFD does, because the cost of carry is already inside the futures price. A leveraged CFD held overnight is financed, and that financing tracks short term dollar interest rates. In the Federal Reserve’s H.15 release dated 1 October 2026 the effective federal funds rate was 3.88 per cent, so a position held for 30 days carries roughly 0.32 per cent of its notional value in financing before any broker markup. That is small for a day trade and material for a position you intend to hold for months.

Choosing without pretending the choice is bigger than it is

Put the three facts side by side and the decision mostly makes itself.

If your account is small enough that a micro futures contract would be more than one or two per cent of it on a sensible stop, you do not have a choice to agonise over. Trade the instrument that lets you size properly, which for most accounts under ten thousand dollars means a CFD account where you can deal in single ounces. The alternative is not “trading futures like a professional”, it is taking a position four or five times larger than your own rules allow, which ends the same way every time.

If your account has grown past the point where micro futures fit inside your risk budget, the futures route becomes genuinely attractive for the structural reasons: an exchange contract, a published specification, a clearing house, no financing charge layered on top. Many people make that move, and it is a reasonable one.

What I would not do is make the move early in order to feel more serious. The size of your position is the only variable in this article that has ever closed an account, and it is the one you control on either venue.

If you are at the stage of asking this question at all, how to trade gold with a small account and how much money to start trading gold are the two pieces that sit either side of it, and what leverage actually is in gold trading covers the mechanism that makes contract size matter so much in the first place.

Frequently asked questions

Is XAUUSD the same as gold futures?

No. XAUUSD is a contract for difference with your broker that settles the change in the gold price. Gold futures are exchange traded contracts in fixed sizes, 100 troy ounces for the full size contract and 10 for the micro, with a clearing house as counterparty. They track the same metal and behave similarly day to day, but the counterparty, the sizing and the cost structure are different.

Which is cheaper to hold overnight?

Futures, usually, because the cost of carry is built into the contract price rather than charged as a nightly financing line. A leveraged CFD is financed at something close to short term dollar rates plus a markup, which was roughly 3.88 per cent a year as a base in the Fed’s H.15 release of 1 October 2026. For a position held a few hours the difference is negligible. For months it is not.

Can I trade micro gold futures with a small account?

You can open one, but opening one and sizing it sensibly are different questions. At a ten dollar an ounce stop the micro contract risks about a hundred dollars, so keeping that inside a one per cent risk budget implies an account near ten thousand dollars. Below that the smallest position available is larger than your rules allow, which is the situation worth avoiding.

Does the 20 to 1 leverage cap apply to me?

It applies to retail clients of firms authorised in the European Union under the ESMA measures linked above. Other jurisdictions set their own limits, some higher and some lower. Check what your broker is authorised under rather than assuming, because the cap also comes bundled with the margin close out rule and negative balance protection, and those matter more than the leverage number itself.

Why is micro gold a fifth of the contracts but only a fortieth of the metal?

Because a micro contract is one tenth the size. A lot of participants holding small contracts produce a high share of the contract count and a low share of the ounces. Both figures are computed from the CFTC report for positions as of 22 September 2026: 412,800 full size and 98,428 micro contracts open.

Which one should a beginner start with?

I will not answer that as a recommendation, because the right answer depends on your account size, your jurisdiction and your broker. What I will say is the test to apply: work out the smallest position the instrument permits, multiply it by the stop distance you would actually use, and check the result against one per cent of your account. If it does not fit, that instrument is not yet for you, whatever anybody calls it.

Where Gold Empire fits

Gold Empire is a free education channel about surviving in gold before trying to grow in it. Everything here is free to read and free to follow on Telegram. The one optional piece is the survival sheet above, a single page on protecting an account. We do not promise profits, we do not publish account results, and we do not believe anyone can tell you what your account will do. If you want the fuller version of the thinking in this article, risk management for gold trading is the piece everything else here hangs off.

About the author. Matthew writes Gold Empire. He traded full size contracts on an account that had no business holding them for most of a year, and the habit that saved him was not a better method but a spreadsheet that worked out the smallest position each venue allowed before he funded it. Nothing in this article is financial advice or a recommendation to trade, and no entry, stop or target discussed should be treated as a signal. Trading leveraged gold carries a substantial risk of loss and is not suitable for everyone.

Somebody sends me a screenshot most weeks with a note that says bearish divergence in gold trading, price is making a higher high while the oscillator is making a lower one, so the top is in. The pattern in the screenshot is usually real. The conclusion drawn from it usually is not. I went and measured every one of these I could find in fifty eight years of gold data, and the honest answer is more interesting than either the people selling the pattern or the people mocking it would like.

Everything below is arithmetic I ran myself on the London afternoon gold auction record, with the assumptions written out so you can disagree with them specifically. There are no gold price levels in this article and nothing here is a recommendation.

What bearish divergence in gold trading actually describes

The pattern has two halves and they have to be read together. Price makes a high, pulls back, then makes a second high that is above the first. Momentum, usually measured with a fourteen period Relative Strength Index, makes a high at the same time as that first price high, then makes a second high that is below the first. Price up, momentum down, in the same stretch of chart.

The reasoning behind it is sound enough as reasoning. RSI measures the size of recent up moves against the size of recent down moves. If price grinds to a new high on smaller daily gains than it used before, the ratio falls even as the price rises. So the pattern is a description of something genuine: the market got to a higher number with less force behind it than last time.

The leap people make is from that description to a forecast. Less force now is taken to mean lower prices soon. That leap is the part I wanted to test, because it is the part that costs money.

How I measured it, and why the method matters more than the result

The price series is the LBMA Gold Price PM, the London afternoon auction print, published by the London Bullion Market Association. It runs from the first of April 1968 to the twenty ninth of September 2026, which is 14,693 trading days. One print per day, so this is an auction record rather than an intraday bar, and a high here means a high in the run of daily fixings.

I defined a swing high as a fixing that is the strict maximum of the eleven day window centred on it, five days either side. That produced 835 swing highs. Then a bearish divergence is a swing high whose price is above the previous swing high while its RSI is below the RSI at that previous swing high, with the two highs between five and sixty trading days apart, and with RSI at the later high at or above 55, which is the usual textbook precondition that the market be at least moderately stretched.

That gives 75 of them in fifty eight years. One every 196 trading days, about 1.29 a year. The first thing worth saying is that this is a rare pattern if you define it strictly, and most of what gets posted in group chats as a divergence would not survive the definition.

The trap that makes the pattern look far better than it is

Here is the part that changed my mind while I was writing the code, and it is the reason I am writing this article at all.

A swing high defined as the maximum of five days either side cannot be identified on the day it happens. By construction, you need the next five days to have printed lower before you can call it a high at all. So if you measure what price did starting from the high itself, the definition has already guaranteed you the first five days of the answer. That is not a finding. That is the test marking its own homework.

When I made that mistake deliberately and measured forward from the high, the pattern looked spectacular. Price was lower ten days later in 85.3 per cent of cases. If I had stopped there, this would be a very different and much more popular article.

Measured properly, from the confirmation day, which is the first day a human being could actually have known the pattern existed, price was lower ten days later in 52.0 per cent of cases. The gap between 85.3 and 52.0 is not a subtlety. It is the whole difference between a tradeable claim and a coin flip, and it is entirely an artefact of when you start the clock.

The size of what gets handed over for free in that window is measurable too. Between the high and the confirmation day, the median case had already fallen 1.92 per cent. A look ahead test gives you that move as a gift. The market does not.

What 75 bearish divergences in gold actually did

Comparing against all days would be a soft test, because a swing high is by definition a local top and local tops tend to be followed by weakness whether or not momentum diverged. So the fair comparison is against other swing highs that were also stretched, RSI at or above 55, and confirmed the same way, but showed no divergence. There are 474 of those.

Ten days after confirmation, price was lower in 52.0 per cent of divergence cases against 48.7 per cent of the control group. Twenty days, 54.7 against 46.8. Sixty days, 41.3 against 44.6, which is to say that at the longest horizon the divergences were followed by slightly fewer declines than the swing highs without them.

So the strongest reading the data supports is this: there is a small effect at twenty days, about eight percentage points over a comparable group, on a sample of 75. That is not nothing. It is also nowhere near enough to carry a decision on its own, and it inverts by sixty days.

Two more numbers finish the picture. In 46.7 per cent of cases the market printed a fixing above the divergence high within twenty days of confirmation, meaning the high the pattern was pointing at did not hold at all. And splitting by decade, the twenty day hit rate runs from 33.3 per cent in the 2020s to 80.0 per cent in the 2000s, on samples of nine to nineteen. A statistic that swings that far between decades on samples that small is not a stable property of the market. It is noise wearing a pattern’s clothing.

Why the pattern survives anyway

Patterns like this persist because of how they are remembered rather than how they perform. A divergence that preceded a fall is memorable, screenshot, and posted. A divergence that preceded a new high is quietly reclassified as a failed signal or as the wrong timeframe, and it does not get posted. Nobody is lying. The sample is just being assembled by memory, and memory is not a random sampler.

The look ahead problem makes it worse, because it is invisible from the inside. If you scroll back through a chart and mark the divergences, your eye finds them at the completed highs, where they are obvious. You are doing the 85.3 per cent version of the test without knowing you are doing it. The pattern feels reliable because you are seeing it under conditions you will never have in real time.

What I think it is honestly good for

I am not telling you to delete the indicator. I am telling you what its output supports.

A bearish divergence is a reasonable description of a market that is climbing with less force than it climbed before. Read as a description, it is accurate and sometimes useful context, in the same way that knowing where the crowd is positioned is context. Read as a forecast of the next twenty days, the evidence is an eight point edge over a comparable group, on 75 samples, that reverses at sixty days and swings between 33 and 80 per cent depending on which decade you happen to be trading in.

The practical consequence is about sizing rather than direction. Nothing in these numbers justifies a bigger position, an earlier entry or a conviction that a level will hold. If an observation moves your size, it needs to be far more reliable than this one is. That is the same test I would apply to market structure or to a liquidity sweep, and most chart observations fail it.

Frequently asked questions

Does a bearish divergence mean gold is about to fall?

On this data, no, not reliably. Measured from the first day the pattern could be seen, price was lower twenty days later in 54.7 per cent of the 75 cases, against 46.8 per cent for comparable stretched highs without a divergence. That is a small edge on a small sample, and it disappears and reverses by sixty days.

Why is your hit rate lower than what other sites report?

Almost certainly because of when the clock starts. If forward returns are measured from the swing high itself, the same data gives 85.3 per cent at ten days rather than 52.0 per cent. A swing high defined with a five day window cannot be known until five days later, so measuring from the high builds the answer into the question.

Would a different RSI setting or timeframe change the result?

It would change the numbers, certainly. It would not change the structural problem, which is that any swing based definition carries a confirmation lag, and any test that ignores that lag will flatter the pattern. If you test a variant, test it from the confirmation bar and compare it against similar highs without divergence rather than against all days.

Is a divergence on a longer timeframe more reliable?

This test used daily auction prints, so I cannot answer for intraday charts from this data, and I would rather say so than guess. What I would point out is that a longer timeframe means a longer confirmation lag in calendar time, so the cost of the delay gets bigger, not smaller.

What about bullish divergence?

Same method, mirrored, and the same caution applies. I have not published those numbers because I have not run them with the same care, and I am not going to assert a result I have not measured.

Where Gold Empire fits

Gold Empire is free to follow. Daily gold analysis with the reasoning written out, the losing days included rather than quietly dropped, plus an optional Kit for people who want the method in one place. There is no promise of profit here, because nobody honest can make one.

If this article stopped you from sizing up on a divergence, it did its job. The habit it belongs to is the unglamorous one: decide what an observation is worth before you let it change your risk. That sits with risk management in gold trading and with reading a gold chart with a clear head, and if you want to know who is writing this, the about page covers it.

About the author

Matthew has spent a long time around gold, most of it learning things the slow way, and now writes them down so other people can learn them the fast way. He has a particular impatience with backtests that start the clock at a moment nobody could have traded, because those are the ones that turn a coin flip into a conviction. Nothing in this article is personal advice, and no entry, stop or target discussed should be treated as a signal.

---

This article is educational content about a chart pattern and about how to test one. It is not investment advice, not a recommendation of any indicator, technique or platform, and not a solicitation to trade. The price series used is the LBMA Gold Price PM published by the London Bullion Market Association, covering 14,693 trading days from April 1968 to September 2026. Every statistic here is my own arithmetic on that series under the definitions stated in the text, a fourteen period Wilder RSI, a five day swing window, an RSI threshold of 55 and a five to sixty day separation between highs; different definitions will give different numbers, which is rather the point. Past behaviour of a pattern is not a forecast of its future behaviour, and no gold price level appears anywhere in this article. Trading carries risk of loss.

Someone in the channel asked me a question that sounds simple and is not. He had found two products at his broker, one called a gold ETF and one called a gold fund, and he wanted to know which one to put his savings into. The question of gold etf vs gold fund which is better gets asked constantly, and almost every answer I have read online compares the wrong things: past performance, ticker popularity, which one the platform pushes on the front page.

None of that is the useful comparison. The useful comparison is structural, and it decides three things that will follow you for as long as you hold the position: what you actually own, what it costs you every year, and what rate your gains get taxed at. Two of those three are knowable in advance with certainty. Let me go through them.

Gold etf vs gold fund which is better starts with what you actually own

The first trap is that neither name tells you what is inside. “Gold fund” is not a legal category, and “gold ETF” describes how a thing trades, not what it holds.

Under those two labels you will find at least three genuinely different products. There are funds that hold physical bullion in a vault. There are funds that hold shares in gold mining companies. And there are funds that hold other funds, or that hold futures contracts rather than metal. A mining fund and a bullion fund can both be sold to you as “gold”, and they behave differently, because a mining company is a business with debt, costs, management and a share price that can fall while the metal rises.

So the first thing to do, before any comparison of fees, is to open the document and find out which of those three you are looking at. That distinction matters more than everything else in this article combined, and it is the one people skip because it takes ten minutes and the fee number takes ten seconds.

What “fund” means legally, and why the wrapper is not decoration

There is a real legal distinction underneath the marketing, and it is worth knowing because it changes what protections apply.

A conventional mutual fund is a registered product. The SEC’s own investor education material describes a mutual fund as an SEC-registered open-end investment company that pools money from many investors, whose portfolio is managed by an SEC-registered investment adviser, and whose investors buy and redeem shares from the fund itself at the next calculated net asset value on any business day.

Read what that gives you: registration, a regulated adviser, and redemption at net asset value once a day. It also tells you what it does not give you, which is the ability to sell at eleven in the morning at a price you can see. You get the day’s NAV, whatever that turns out to be.

A bullion-backed exchange traded product is often organised differently, commonly as a trust that holds metal, rather than as a registered investment company of that kind. It trades on an exchange all day like a share, so you can deal intraday at a visible price, and you are buying from another market participant rather than from the fund. That is a genuine convenience. It is also a different legal object with a different rulebook, and you should not assume the protections are identical because the marketing language is.

The practical instruction here is boring and I am going to give it anyway: find the prospectus, find the section that says how the thing is organised, and read that section. It is usually one page. It is the highest return ten minutes available to a person at this stage of the decision.

The tax difference that almost nobody checks first

Now the part that surprises people, and the reason I think this question deserves a whole article rather than a one line answer.

In the United States, long term capital gains for most assets are capped at a top rate of 20 per cent. But the IRS treats collectibles differently. Its guidance on capital gains states plainly that net capital gains from selling collectibles are taxed at a maximum 28 per cent rate.

And gold is a collectible for this purpose. IRS Publication 550 defines a collectibles gain or loss as gain or loss from the sale or trade of a work of art, rug, antique, metal (such as gold, silver, and platinum bullion), gem, stamp, coin, or alcoholic beverage held more than 1 year. The same passage adds the sentence that matters most here: collectibles gain includes gain from the sale of an interest in a partnership, S corporation, or trust due to unrealized appreciation of collectibles.

Hold those two together. If the product you own is a trust whose appreciation comes from bullion, the gain can be a collectibles gain, taxed at up to 28 per cent rather than up to 20 per cent. On 10,000 USD of gain that is the difference between keeping 8,000 USD and keeping 7,200 USD, or 800 USD, purely because of how the wrapper is built.

I want to be careful about how far I take that. Your own situation, your country, your account type and the specific structure of your fund all change the answer, and a fund held inside a tax sheltered retirement account is a different conversation entirely. I am not a tax adviser and this is not tax advice. What I am telling you is that the wrapper has a tax consequence, that the consequence is large enough to notice, and that it is the sort of thing worth asking a qualified person about before you buy rather than in April afterwards.

What the annual fee does while you are not looking

The third structural difference is the annual fee, and this is the one you can calculate exactly.

A fund charges a percentage of assets every year, whether the year was good or bad. Bullion exchange traded products tend to sit at the lower end, and actively managed funds and funds that hold other funds tend to sit higher, because the second layer of management gets paid too.

Take 10,000 USD, assume an identical 6 per cent gross annual return for both, and change nothing but the fee. After 20 years the 0.40 per cent product is worth 29,736 USD and the 1.00 per cent product is worth 26,533 USD. The gap is 3,203 USD, which is 10.8 per cent of the cheaper result. Push the fee to 1.50 per cent and the gap becomes 5,619 USD, or 18.9 per cent.

Those are assumptions, stated openly: same gross return, fee deducted annually, no trading costs, no tax in that particular calculation, and a 6 per cent figure chosen to make the arithmetic comparable rather than as any forecast of what gold will do. Nobody knows what gold will do. The fee, by contrast, is published in advance and is one of the very few numbers in this business you can know before you commit.

That is the reason I keep returning to costs in everything I write. You cannot control the return. You can control the fee and the tax treatment, and over twenty years those two do more visible damage than most of the things people worry about instead.

So which one is better

Neither, as a general answer, and I would be suspicious of anyone who gives you one without asking what you are trying to do.

What I can offer is the order to think in. Decide what exposure you want first, metal or mining companies, because that is the largest difference and it is not a cost question. Then check the annual fee, because it is knowable and it compounds. Then check the structure and its tax consequence, because an 8 percentage point difference in rate is not a detail. Then, last, consider whether intraday dealing actually matters to you, which for someone buying to hold for years it usually does not.

And if what you actually want is the metal itself rather than a claim on it, that is a third road with its own costs, covered in gold ETF vs physical gold returns. If you are weighing metal against the companies that dig it up, is it better to buy physical gold or gold stocks takes that comparison apart.

One thing this article is not about, and I want to say it clearly: none of this is a view on whether you should own gold at all, or when. It is about what happens to the money once you have decided. Whatever you hold it in, size it as part of a plan rather than as a reaction, which is the argument in risk management in gold trading.

Frequently asked questions

Is a gold ETF always cheaper than a gold fund?
Usually but not reliably, and “usually” is not a good enough basis to skip checking. Bullion exchange traded products tend to sit at the low end of the fee range because they are doing something mechanical, holding metal, while an actively managed fund is paying people to make decisions and a fund that holds other funds is paying two layers. But fee ranges overlap, and a cheap wrapper around the wrong exposure is not a bargain. Look up the actual published figure for the actual product rather than reasoning from the category name.

Does the 28 per cent collectibles rate apply to every gold product?
No, and this is exactly why the structure matters. The rate attaches to collectibles gain, which the IRS defines to include gold bullion and to include gain from selling an interest in a trust due to unrealized appreciation of collectibles. A fund holding mining company shares is holding equities, not metal, so it is not the same case. Account type changes it again. The honest answer is that it depends on the specific product and on your situation, which is the whole argument for reading the prospectus and asking a qualified person in your country.

Where did the fee numbers in the chart come from?
They are my own arithmetic, not quoted from any provider. I took 10,000 USD, applied a 6 per cent gross annual return, deducted the fee annually, and compounded for 20 years at three fee levels. The 6 per cent is a comparison device, not a prediction, and the point survives whatever number you substitute: the gap between fee levels is a fixed share of the result and grows with time. Put your own product’s published fee into the same calculation and you will get a figure that means something for your case.

Can I just hold the metal instead?
You can, and plenty of people should consider it, but understand that you are swapping one set of costs for another rather than escaping costs. Physical metal brings a dealer spread when you buy and again when you sell, storage or insurance if you do not want it in a drawer, and the practical problem of verifying and selling it later. It removes the annual management fee and it removes counterparty questions. Which set of costs suits you depends on size, timeframe and how much practical hassle you are willing to absorb.

Does any of this tell me whether gold will go up?
Not even slightly, and I would be wary of any page that mixes the two questions. Everything here is about the container, not the contents. Choosing a cheaper wrapper does not make gold rise, and choosing the most expensive one does not make it fall. The reason to get the container right is that it is the part of the outcome you can actually determine in advance, which is rare enough to be worth the ten minutes.

What Gold Empire actually does

Gold Empire is a free Telegram channel and a free library of market notes about gold, written for people who want to survive their first few years in this market rather than get rich in them. Everything on the site is free and there is no profit promise anywhere on it, because nobody can make one honestly. The model is simple: follow along for free, and take the free survival sheet if it is useful to you.

About the author. Matthew writes the Gold Empire market notes and has spent most of his time in this market learning what not to do. He is not a licensed adviser and does not want to be one, and he paid a fee for years without once checking what it compounded to. His interest is in the part of this business nobody posts screenshots of: staying in the game long enough for a method to matter.

Disclaimer: This article is general educational content about the structure and cost of investment products. It is not financial advice, not tax advice, not a recommendation, and not a solicitation to buy or sell anything. All cost figures are illustrations computed from stated assumptions and are not quoted from any provider. Tax treatment depends on your individual circumstances and on the jurisdiction you are in, and it can change. Trading and investing carry a risk of losing money, and no entry, stop or target discussed should be treated as a signal. Consider your own circumstances and seek independent regulated advice if you need it.

The question is it better to buy physical gold or gold stocks gets asked as though the two things are the same asset in different packaging, one of them simply more convenient than the other. They are not. A bar of gold and a share in a gold mining company are different kinds of thing entirely, and the difference is not convenience or storage. It is that one of them has a cost line and the other does not.

That single structural fact explains almost everything people find surprising about gold equities, including the part that catches out most newcomers, which is that a gold share can fall on a day when gold rises. If you understand where the cost line sits, that stops being a mystery and becomes arithmetic.

What You Actually Own in Each Case

Hold physical gold and you own a quantity of metal. Its value moves one for one with the price of that metal, less whatever you paid to acquire and store it. There is no management, no payroll, no diesel bill and no mine that might flood. There is also nothing that can turn a 10 per cent move in gold into a 30 per cent move in your position. What the metal does, you get.

Hold a gold mining share and you own a slice of a business whose product happens to be gold. That business digs metal out of the ground at a cost, sells it at the prevailing price, and keeps the difference. You own the difference, not the metal. You also own everything else that comes with a company: the debt, the jurisdiction the mine sits in, the quality of the management, the share count, and the possibility that any of those change without the gold price moving at all.

Gold Empire is mostly about leveraged trading rather than long term holding, so I want to be clear why this matters here. The reason is that traders who move from trading gold to owning gold equities usually assume they are taking less risk, because equities feel tamer than a leveraged position. On the numbers below, that assumption is frequently wrong, and it is wrong in a way that only shows up once you are holding it.

The Cost Line, and Why the Industry Publishes It

Mining companies report a standardised figure called all-in sustaining cost, usually shortened to AISC. It is designed, in the World Gold Council’s own words, to better reflect the full cost of keeping the mine in business, which means it includes not just digging the metal out but the ongoing capital needed to keep doing so next year.

The World Gold Council publishes a quarterly series of the global average AISC of gold production, running from 2012 and updated alongside the Metals Focus Gold Mine Cost Service. You can see the all-in sustaining cost data on Goldhub directly, and the numbers in that series are worth reading slowly.

In the first quarter of 2012, the global average AISC was US$926.78 an ounce. In the first quarter of 2026, the most recent quarter in the series as of 31 March 2026, it was US$1,785.49 an ounce. That is a rise of 92.7 per cent over fourteen years, a compound rate of 4.80 per cent a year. Narrow the window and it is sharper still: the first quarter of 2021 sat at US$1,058.59, so the cost of producing an ounce climbed 68.7 per cent in five years.

Note what this figure is and is not. It is a cost of production, not a gold price, and nothing in this article quotes a gold price. But it is the number standing between the metal price and the shareholder, and it has been moving, upward, for over a decade.

Is It Better to Buy Physical Gold or Gold Stocks for a 10 Per Cent Move?

Now the arithmetic, and it needs no price level at all, only the relationship between the selling price and the cost.

Suppose a producer sells its metal at some multiple of its all-in sustaining cost. Call that multiple m. Its gross margin per ounce is the price minus the cost, which is (m minus 1) times the cost. Now move the metal price by 10 per cent and ask what happens to that margin. The answer falls out of the formula: the margin changes by 0.1m divided by (m minus 1).

Run it at four assumptions:

• At a price 1.15 times cost, a margin of 13 per cent on revenue, a 10 per cent move in the metal moves the gross margin 76.7 per cent. An amplification of 7.7 times.
• At 1.25 times cost, a 20 per cent margin, the move is 50 per cent. Amplification of 5 times.
• At 1.50 times cost, a 33 per cent margin, the move is 30 per cent. Amplification of 3 times.
• At 2.00 times cost, a 50 per cent margin, the move is 20 per cent. Amplification of 2 times.

Two things are worth saying plainly about that table.

The first is that this is exactly what leverage looks like, arriving through the income statement rather than through a margin account. A trader who would never open a five times leveraged position on gold may cheerfully buy a miner priced at 1.25 times its production cost, which delivers a similar amplification of the metal’s move into the thing they actually own. Nobody calls it leverage because no broker is lending anything. The mechanism is different, the arithmetic is not.

The second is that it runs both ways, and this is where people get hurt. At 1.25 times cost, a 10 per cent fall in the metal cuts the gross margin by 50 per cent. The same producer, the same mine, the same tonnage, half the margin. The amplification is not a feature that only operates in your favour, and the tighter the margin, the more brutal the effect. A producer at 1.15 times cost loses 76.7 per cent of its gross margin on a 10 per cent decline in gold.

The Costs Do Not Stand Still Either

Here is the part the amplification table hides, because it treats cost as fixed while the price moves. Over any real holding period, it is not.

Go back to the AISC series. A mine operating at the Q1 2012 global average that now operates at the Q1 2026 average would need its selling price to be 1.93 times what it was, just to hold the same absolute margin per ounce. Not to grow the margin. To stand still.

The person holding physical metal has no equivalent line in their accounts. Their ounce did not get more expensive to produce, because they are not producing it. This is the structural asymmetry at the heart of the comparison, and it is why gold equities and gold do not track each other over long periods even though they are driven by the same commodity.

Supply gives you a sense of how slowly the other side moves. In the second quarter of 2026, according to the World Gold Council’s Gold Demand Trends, mine production was 965.6 tonnes, up 2 per cent year on year. Mine output barely moves. A company cannot dig its way out of a cost problem at any speed that would help you.

The Risks That Have Nothing to Do With Gold

Everything above assumes the only thing that changes is the metal price and the cost of getting it out. A share carries more than that.

It carries dilution, because a company that needs capital can issue more shares, and your claim on the margin gets smaller without the margin changing. It carries jurisdiction, because a mine sits in a specific country with specific tax and permitting rules that can be rewritten. It carries operational risk, because a single mine can flood, a grade can disappoint, a strike can stop production. It carries the equity market itself, which means that on the day everything falls together, your gold exposure can fall with it because it is priced in a stock market rather than a metals market.

None of these are arguments against owning gold equities. They are arguments against owning them while believing you own gold. Those are different positions with different failure modes, and the failure mode is the thing I care about.

So Which One

The honest answer is that they do different jobs, and the question is what job you want done.

If the job is to hold a store of value that moves with the metal and cannot be diluted, mismanaged or nationalised out from under you, that is what physical gold does, and paying storage is the price of that property. If the job is to take a geared position on the gold price with the gearing supplied by a business rather than a broker, that is what a gold equity does, and the amplification table above is the specification you are buying, in both directions.

What neither of them is, is the other one with better logistics. The most expensive version of this decision is the one made by somebody who wanted the first job done and bought the second thing because it was easier to buy.

Frequently Asked Questions

Is it better to buy physical gold or gold stocks if I want exposure to the gold price?
Physical tracks the metal one for one. A gold equity gives you an amplified, two way version of the same move, and the amplification depends on how far the selling price sits above production cost. At 1.25 times cost, a 10 per cent move in gold becomes a 50 per cent move in gross margin, upward or downward.

Why do gold stocks sometimes fall when gold rises?
Because you own the margin, not the metal, and the margin has two sides. Costs can rise faster than the price, a company can dilute, a mine can underperform, and the whole equity market can drag a share down regardless of what the commodity did that day.

Are gold miners a form of leverage?
In effect, yes, though nobody is lending you anything. The gearing comes from the gap between price and production cost rather than from a margin account. The arithmetic looks very similar, and it is worth treating the position with the same caution you would give any geared exposure.

Does a rising all-in sustaining cost mean miners are a bad investment?
Not on its own, because the selling price moves too. It does mean the margin is a moving target rather than a fixed property of the business, and that a company standing still on cost control is falling behind. The global average AISC rose 92.7 per cent between the first quarter of 2012 and the first quarter of 2026.

Where did the numbers in this article come from?
The all-in sustaining cost figures are from the World Gold Council’s quarterly AISC series on Goldhub, as at 31 March 2026, and the mine production figure is from its Gold Demand Trends for the second quarter of 2026. Both are linked above. The amplification percentages are the formula 0.1m divided by m minus 1, evaluated at four stated assumptions about how far the selling price sits above cost. They illustrate a relationship and are not a forecast for any company. No gold price level is quoted anywhere in this article.

Where Gold Empire Fits

Gold Empire is free to follow. Daily XAU/USD analysis on Telegram, with the reasoning stated before the trade rather than after it, losing days included. There is a free survival sheet for anyone who wants the capital protection material in one place, and an optional Kit for people who want the full method. Nothing here promises a profit and nothing here is a signal service.

The pillar this sits under is risk management for gold trading, the wrapper comparison is in gold ETF vs physical gold returns, and the failure mode of a geared position is covered in gold CFD vs physical gold. More about who writes this is on the about page.

About the author. Matthew writes Gold Empire and spends most of his attention on the boring mechanics that decide whether an account is still open in two years. He has a long standing habit of reading cost disclosures before price charts, on the grounds that the cost line is the part nobody puts in the marketing. Nothing in this article is a recommendation to trade, and no entry, stop or target discussed should be treated as a signal.

---

Disclaimer: This article is general educational content about the structural differences between holding physical gold and holding shares in gold mining companies. It is not financial advice, not investment advice and not a recommendation to buy, sell or hold any instrument, product or company. Production costs, company accounts and tax rules change, so verify current details with primary sources before acting. Figures are taken from the stated public sources or are simple arithmetic on clearly stated assumptions, not predictions or performance claims. Trading leveraged products carries a substantial risk of loss and most retail accounts lose money.

Open a gold futures vs XAUUSD chart side by side and the first thing you notice is that they look almost identical. The second thing, if you look for another minute, is that they are not. The shapes agree, the turning points agree, but the candles do not line up, the sessions start and stop in different places, and every so often one of them has a gap that the other does not. Traders usually assume one of the two is wrong. Neither is. They are two different instruments that happen to track the same metal.

This matters more than a chart curiosity, because the difference decides what you are actually holding, what it costs you to hold it, and who you are holding it with. I want to walk through the mechanics in the order they affect your account.

Two Markets, One Metal

The futures side lives on an exchange. COMEX lists a gold contract and that contract is a standardised, fixed lump of metal. You can see exactly how big it is in a source neither your broker nor I control: the US Commodity Futures Trading Commission publishes a weekly Commitments of Traders report, and the COMEX section of it labels each market with its contract size on the line above the data.

In the report covering positions as of 15 September 2026, the entry reads GOLD, COMMODITY EXCHANGE INC., contracts of 100 troy ounces, with open interest of 409,899 contracts. Directly beneath it sits MICRO GOLD, contracts of 10 troy ounces, with open interest of 90,365. You can read the COMEX Commitments of Traders report yourself, and it will have moved on by the time you do, which is rather the point of using it.

The spot side lives somewhere else entirely. XAUUSD as you see it in a retail terminal is not an exchange product. It is a contract for difference or a spot position quoted by your broker, priced off the over the counter loco London market, where gold has traded between banks for a very long time without any central exchange at all.

The Arithmetic That Makes the Two Comparable

Here is the part that makes the comparison concrete, and it needs no price at all.

One COMEX gold contract is 100 troy ounces. One standard lot of XAUUSD is also 100 troy ounces. So a single standard lot in your retail terminal carries the same quantity of metal as one full exchange contract. That is a fact worth sitting with if you have been treating a standard lot as a normal starting size.

Work down the scale. One standard lot equals ten micro gold contracts. A 0.01 lot, the smallest size many brokers allow, is one troy ounce, which is a tenth of a single micro contract. The retail market slices the metal far finer than the exchange does, and that is genuinely the main practical advantage of the spot side.

The open interest figures show how that plays out. Micro contracts are 18.1 per cent of all open COMEX gold contracts, which sounds like a substantial share of the market. Converted to metal, using the contract sizes printed in the same report, the full contracts account for 40,989,900 ounces and the micros for 903,650. The micro contract is therefore 2.2 per cent of the gold. Counting contracts and counting metal are two different jobs, and confusing them is how people talk themselves into believing a market is smaller or more retail than it is.

Why the Charts Do Not Line Up

Four mechanical differences produce almost every visible discrepancy between the two charts.

The futures chart is stitched together. A futures contract expires. The chart you are looking at, if it runs back further than a few months, is a continuous series assembled from a sequence of contracts, each handed over to the next at some roll date. Every handover can leave a step in the series that never happened to anyone holding a position. Spot XAUUSD has no expiry, so its chart has no rolls and no stitches.

The cost of time is expressed differently. Because a futures contract has a delivery date, the cost of holding metal over time is baked into the price of the contract itself. On the spot side the same economics arrive as a financing charge applied to your position, which is why an overnight XAUUSD position is debited or credited while the chart price carries no equivalent adjustment. We cover that mechanism in what is swap in gold trading.

Settlement is a different animal. Spot loco London settles two good UK business days after the trade, per ICE Benchmark Administration, which administers the LBMA Gold Price. A futures contract settles at its delivery date, potentially months away. Two instruments with different settlement horizons are not obliged to print the same price at the same moment.

The clocks are different. The exchange has a defined session with a defined daily break. The over the counter market has trading hours defined by banks in different time zones handing the book around. On top of that, the LBMA Gold Price runs auctions twice each business day, at 10:30 and 15:00 London time, with price formation in US dollars, as described by ICE Benchmark Administration and by the LBMA itself. That is ten benchmark auction moments in a normal working week which sit inside the over the counter market and have no equivalent on the futures chart.

Why Most Retail Accounts End Up on the Spot Side Anyway

Given that the exchange product is centrally cleared and transparently specified, it is fair to ask why almost every retail gold trader ends up on the spot side instead. There are three honest reasons and none of them is that one product is better.

The first is granularity, which we have already covered. The exchange sells metal in lumps of 100 ounces, or 10 for the micro. If your risk budget on a single idea is small, the smallest exchange lump may already be too large to be sized sensibly, whereas a 0.01 lot of one troy ounce almost certainly is not.

The second is access. A futures account is a different product from a retail spot account, with different paperwork, different minimums and a different relationship to margin. Plenty of traders never look at the futures side simply because the spot account was the one in front of them.

The third is the absence of expiry. A spot position can be held indefinitely, subject to financing, while a futures position has a date on it and must be closed or rolled. For somebody holding a view over weeks rather than hours, not having to manage a roll is a real convenience, even though the cost of time has not disappeared, it has only changed its address.

The flip side of those three conveniences is the counterparty point below, and it is a genuine trade, not a free lunch. Exchange clearing is something you give up in exchange for granularity and ease of access.

What Your Gold Futures vs XAUUSD Chart Comparison Should Change

Three practical consequences, in order of how much money they touch.

First, the counterparty question. A COMEX position is cleared through the exchange clearing house. A retail XAUUSD position is a contract with your broker, which means your broker’s solvency and conduct are part of your risk, not an afterthought. That is not an argument against the spot side. It is an argument for choosing the firm carefully, which is what choosing a broker for gold trading is about.

Second, the sizing question. If one standard lot is one full exchange contract, then the difference between 0.01 and 1.00 in the volume box is a factor of one hundred in metal. The fine granularity of the retail market is a genuine gift for risk control, and it is also the thing that lets somebody place an exchange sized position by accident in a terminal that makes it feel like a small number. The mechanics of translating that into an actual risk figure are in how to calculate lot size for gold forex.

Third, the data question. If you analyse on one chart and execute on the other, your levels are drawn from a series that your fills will not exactly respect. The gaps you see on a stitched futures chart are not gaps your spot position can trade through, and the session boundaries differ. Analyse on the instrument you trade, or at minimum know which of the two you are looking at.

Frequently Asked Questions

Is XAUUSD the same as gold futures?
No. XAUUSD in a retail terminal is a spot or contract for difference position with your broker, priced off the over the counter loco London market. Gold futures are standardised exchange contracts with expiry dates, cleared centrally. They track the same metal closely, which is why the charts look alike, but the instrument, the counterparty, the expiry and the way the cost of time is charged are all different.

Which chart shows the real gold price?
Both show a real price for a real instrument, and neither is the single official price of gold. If you want a published benchmark rather than a live quote, the LBMA Gold Price is set by auction twice each business day at 10:30 and 15:00 London time and is published by ICE Benchmark Administration.

Why does my broker’s gold chart differ from the futures chart I see elsewhere?
Usually some combination of four things: the futures series is stitched across contract rolls, the two instruments have different settlement horizons, the cost of holding time sits inside the futures price but arrives as a financing charge on the spot side, and the trading sessions do not share the same boundaries. Quote source differences add a little more on top.

How much gold is in one standard lot?
One hundred troy ounces, the same quantity as one full COMEX gold contract as labelled in the CFTC report. One micro gold contract is ten troy ounces, and a 0.01 lot is one troy ounce.

Where did the numbers in this article come from?
The contract sizes and the open interest figures are taken directly from the CFTC Commitments of Traders report for COMEX, futures only, positions as of 15 September 2026, and that report is linked above. The percentage shares are simple multiplication of those two published figures, open interest multiplied by the contract size printed on the same line. The auction times and the settlement convention come from ICE Benchmark Administration. No price level is quoted anywhere in this article.

Where Gold Empire Fits

Gold Empire is free to follow. Daily XAU/USD analysis on Telegram, with the reasoning stated before the trade rather than after it, losing days included. There is a free survival sheet for anyone who wants the capital protection material in one place, and an optional Kit for people who want the full method. Nothing here promises a profit and nothing here is a signal service.

The pillar this sits under is risk management for gold trading, the definition work is in what is spot gold meaning, and the leverage that turns a hundred ounce position into a small looking number is covered in what is leverage in gold trading. More about who writes this is on the about page.

About the author. Matthew writes Gold Empire and spends most of his attention on the boring mechanics that decide whether an account is still open in two years. He has a long standing habit of reading contract specifications before charts, which makes him poor company at trading meetups and has saved him more money than any indicator. Nothing in this article is a recommendation to trade, and no entry, stop or target discussed should be treated as a signal.

---

Disclaimer: This article is general educational content about market structure, contract specifications and instrument differences in gold trading. It is not financial advice, not investment advice and not a recommendation to buy, sell or hold any instrument, product or firm. Contract specifications and broker terms change, so verify current details with the exchange and with your own broker before acting. Figures are taken from the stated public sources or are simple arithmetic on them, not predictions or performance claims. Trading leveraged products carries a substantial risk of loss and most retail accounts lose money.

Sooner or later someone in a chat group mentions options, usually in the same breath as the words limited risk, and the question arrives: how to trade in gold options, and is it safer than what I am doing now. It is a fair question and it has a real answer, but the answer is not on the buy side of the screen. It is in the arithmetic of what the contract charges you before you are right about anything.

Here is the short version so nobody has to read three thousand words to understand the shape of it. You pay a fee up front for the right, not the obligation, to trade gold at a fixed price by a fixed date. If gold does not move far enough in your direction before that date, the fee is gone. Your maximum loss is known on the day you buy. So is the fact that the most likely outcome is losing all of it.

The part worth staying for is the other side of the trade, the one where somebody sells you that right. That side has no known maximum loss at all, and it is the side that retail accounts wander into by accident when a platform makes it look like collecting income.

How to Trade in Gold Options, the Mechanical Part

Start with the vocabulary, taken from the regulator rather than from a course, because the definitions are precise and the marketing versions are not. The US Commodity Futures Trading Commission publishes a full glossary of these terms, and it is worth reading in the original.

• The premium is, in the CFTC’s words, “the payment an option buyer makes to the option writer for granting an option contract.” It is paid at the start and it is not refundable.
• The exercise price, also called the strike price, is “the price, specified in the option contract, at which the underlying futures contract, security, or commodity will move from seller to buyer.”
• The expiration date is “the date on which an option contract automatically expires.” After it the contract is worth its intrinsic value, or nothing.
• The option writer is “the person who originates an option contract by promising to perform a certain obligation in return for the price or premium of the option.”

Note the asymmetry in that last definition. The buyer has a right. The writer has a promise to perform.

From those four pieces you get four possible positions. Buy a call, the right to buy gold at the strike, which pays if gold rises far enough. Buy a put, the right to sell at the strike, which pays if gold falls far enough. Or write either of those, collect the premium, and take on the promise.

Where they trade matters more than most beginners realise. Exchange-listed gold options are options on gold futures contracts, cleared by a central counterparty, with published contract sizes and standardised terms. Everything else is a bilateral arrangement with whoever is on the other side of your account, which makes the counterparty part of the risk. Find out which of the two you are being offered before anything else.

The Price of the Ticket, Measured

An option’s price is not arbitrary. You can get close to it with public data and a page of arithmetic, which means you can check whether what you are being quoted is roughly sane before you pay it.

The dominant input is volatility, so start there. Taking the London Bullion Market Association’s published gold benchmark history and measuring the daily variation of the benchmark from January 2000 to September 2026, across 6,674 observations, gold’s realised annualised volatility works out at 17.42 per cent. Over the most recent ten years it was 16.06 per cent, over five years 17.98 per cent, over three years 20.27 per cent, and over the last twelve months 27.66 per cent. The metal has been considerably livelier lately than its own long-run average.

Feeding the long-run 17.42 per cent figure into the standard option pricing formula, with the assumptions written out below, gives the cost of a call as a percentage of the gold price:

• A call struck at today’s price, expiring in 30 days, costs about 1.99 per cent of the gold price. At 90 days, 3.45 per cent. At a year, 6.94 per cent.
• A call struck 5 per cent above today’s price costs about 0.45 per cent at 30 days, 1.58 per cent at 90 days, and 4.89 per cent at a year.
• A call struck 10 per cent above costs about 0.06 per cent at 30 days and 0.62 per cent at 90 days.

The assumptions, stated openly because a number without its assumptions is a decoration: the Black-Scholes formula, a zero interest rate, no storage or lease costs, constant volatility, and price movements that follow a log-normal distribution. Real option markets have none of those things exactly. Real quotes also carry a bid-ask spread, and the far out-of-the-money contracts in the list above have the widest ones relative to their value. Treat these as the shape of the cost, not as a price list.

The Number That Decides Everything

Now combine the two halves, because this is the calculation almost nobody does before buying their first option.

A call only pays at expiry if gold finishes above the strike. But you did not get the strike for free, you paid a premium for it. So the price at which you actually begin to make money is the strike plus the premium, and the move required is both of those added together.

On the 90 day contracts priced above:

• Struck at today’s price: premium 3.45 per cent, so gold must rise 3.45 per cent just to get your money back. Probability of finishing above that, on the same model: 33.17 per cent.
• Struck 3 per cent above: premium 2.20 per cent, so the required rise is 5.20 per cent. Probability: 26.44 per cent.
• Struck 5 per cent above: premium 1.58 per cent, required rise 6.58 per cent. Probability: 21.76 per cent.
• Struck 10 per cent above: premium 0.62 per cent, required rise 10.62 per cent. Probability: 11.31 per cent.

Read the shape rather than the individual numbers. The cheap-looking contracts are cheap because they are unlikely to pay. The far out-of-the-money call, the one usually described as a lottery ticket, costs a tenth as much as the one struck at today’s price and needs three times the move. That is not a bargain, that is the price being approximately right.

Note also what this does to a trader used to spot gold. On a normal position you can be roughly right about direction and slightly wrong about timing and still end up fine. On a bought option you can be exactly right about direction, watch gold rise two per cent, and lose every penny because the contract expired on a Friday and the move happened the following Tuesday.

Time Is the Other Side of the Trade

The CFTC glossary calls it time decay: “the tendency of an option to decline in value as the expiration date approaches, especially if the price of the underlying instrument is exhibiting low volatility.”

Put numbers on it. Take that 90 day call struck at today’s price, and hold gold perfectly still. What is left of your premium as the clock runs?

• With 60 days to go: 81.66 per cent of what you paid.
• With 45 days to go, the halfway point: 70.7 per cent.
• With 30 days to go: 57.75 per cent.
• With 7 days to go: 27.90 per cent.
• With 1 day to go: 10.54 per cent.

Half the calendar gone, roughly seventy per cent of the value left, and then it falls off a cliff. Nothing has to go wrong for that to happen. The market does not have to move against you. Standing still is enough, and standing still is what markets mostly do.

The Side That Can Take More Than You Put Up

Everything above is the buyer’s side, where the worst case is losing the premium. Now the other side, because this is where accounts actually die.

The CFTC defines a naked option as “the sale of a call or put option without holding an equal and opposite position in the underlying instrument.” You collect the premium up front. In exchange you have promised to perform, and if gold runs, your obligation grows with it, without a ceiling.

Here is that in numbers, using the same model and the same public price history. Sell a 30 day call struck 5 per cent above the market and you collect about 0.45 per cent of the gold price. Now look at what gold has actually done. Across every 30 calendar day window from 2000 to September 2026, the largest rise was 23.75 per cent, from 30 December 2025 to 29 January 2026, with 22.26 per cent in 2006 and 21.57 per cent during September and October 2008 close behind.

Meet the biggest of those while short that call and you owe 18.75 per cent of the gold price against the 0.45 per cent you collected. That is 42.1 times your income on a single contract, and it did not happen in the distant past. It happened this year.

And it is not a freak. Of all 6,679 thirty day windows in the sample, 18.16 per cent saw gold rise more than 5 per cent. Nearly one in five. The strategy that looks like collecting rent works most of the time, which is precisely why people scale it up, and then one window in the tail removes more than every premium collected before it.

If you take nothing else from this article, take this: buying an option and writing one are not two variations of the same trade. They are opposite risk shapes wearing the same name. A platform that offers both behind the same button is not doing you a favour.

What This Does Not Say

It does not quote a price for any gold option. Every figure here is a model output from stated assumptions, produced to show the shape of the cost, and real quotes will differ, sometimes materially.

It does not say the model is right. Black-Scholes with constant volatility and zero rates is a simplification. Real markets price a volatility skew, real gold has fatter tails than the log-normal distribution, and those fat tails make the sold-option case worse than the arithmetic above, not better. Where the model is wrong, it is wrong in the direction of understating the danger of writing.

It does not predict anything, and it does not tell you to trade options or not to. Measuring the last twenty-six years describes the past, and the 23.75 per cent window is a reminder that the historical maximum was set recently rather than a forecast. What this article gives you is what the contract charges before you are right, which is information you are entitled to have before you decide.

Five Questions to Answer Before You Trade a Gold Option

None of these are about direction. They are the questions that decide whether a loss is survivable.

• Is it exchange-listed and centrally cleared, or is my counterparty the firm holding my money? If it is the second, the counterparty is part of the risk.
• What is one contract worth in my currency? Not in per cent, in money. If you cannot state the figure, you cannot size the position.
• What is the bid-ask spread as a share of the premium? On a contract costing 0.62 per cent of the gold price, a spread worth a tenth of the premium is a large cost before anything else happens.
• Am I buying or writing, and what is my maximum loss? If you are writing, there is no maximum, and you need to know how your broker will close you out. Leverage in gold trading covers why that close-out arrives sooner than people expect.
• Does this instrument fix a problem I actually have? Accounts rarely blow up for lack of a second instrument. They do so for the reasons in why you keep blowing your trading account, none of which are solved by adding complexity.

Frequently Asked Questions

How to trade in gold options as a complete beginner?
The honest answer is that this is the wrong first instrument. An option adds a second variable, time, to a trade you are already trying to get right on direction and size. If direction and size are not yet reliable, adding a clock does not help.

Are gold options safer than spot gold or CFDs?
A bought option has a known maximum loss, which is genuinely different from a leveraged spot position. A written option has no maximum loss at all. So the instrument is not safer or riskier; the side you take is what decides that.

Why did my option lose money when gold went my way?
Almost always time decay, sometimes a fall in priced volatility. Gold has to move far enough, and soon enough. The 90 day at-the-money contract above needs a 3.45 per cent rise simply to return the premium.

Is selling options a reliable income strategy?
It produces small regular receipts and rare very large payments out. On this data, 18.16 per cent of 30 day windows saw gold rise more than 5 per cent, and the worst rose 23.75 per cent, which is 42.1 times the premium a 5 per cent out-of-the-money writer would have collected. Frequency of winning is not the same as expectancy.

Where did the numbers in this article come from?
Volatility, the 30 day windows and the historical maximum rise were calculated from the published LBMA gold auction history linked below, downloaded before publication. Premiums, break-even moves, probabilities and decay were computed with the Black-Scholes formula under the assumptions stated in the text. The definitions are quoted from the CFTC glossary linked above. None of it is a price quote, a forecast or a performance claim.

Where Gold Empire Fits

Gold Empire is free to follow. Daily XAU/USD analysis on Telegram, with the reasoning stated before the trade rather than after it, losing days included. There is a free survival sheet for anyone who wants the capital protection material in one place, and an optional Kit for people who want the full method. Nothing here promises a profit and nothing here is a signal service.

A new instrument is a risk decision before it is anything else. Risk management for gold trading is the pillar this article sits under, how much to risk per trade is the arithmetic that decides whether any of this is survivable, and the difference between gold and gold futures covers the instrument these options are usually written on. The price history comes from the London Bullion Market Association price record. More about who writes this is on the about page.

About the author. Matthew writes Gold Empire and spends most of his attention on the boring mechanics that decide whether an account is still open in two years. He has watched more than one trader discover that the side of an option trade they were on was not the side they thought. Nothing in this article is a recommendation to trade, and no entry, stop or target discussed should be treated as a signal.

---

Disclaimer: This article is general educational content about derivative instruments, option pricing mechanics and risk in retail trading. It is not financial advice, not investment advice and not a recommendation to buy, sell, write or use any product, contract, platform or firm. Availability of gold options, contract specifications, tax treatment and regulatory protections vary by jurisdiction and change over time, and current terms should be checked with your own broker and with the sources linked in the text. Options can expire worthless and writing options can produce losses greater than the premium received and greater than the capital in the account. Trading leveraged products carries a substantial risk of loss and most retail accounts lose money. Figures are calculations from the stated public data and stated assumptions, not predictions or performance claims.