Market Clues Handbook

How We Read the COT Report

The COT report is one of those datasets that looks simple until you start working with it. Every week, the CFTC publishes a snapshot of how different groups of futures traders are positioned. You can see long and short positions, spreading activity and open interest, broken down by trader category.

The numbers themselves are easy to find. The more interesting part is understanding what changed and what may be behind that change.

At Market Clues, we do not treat the COT report as a shortcut to a bullish or bearish conclusion. We use it to understand participation behind the price chart. Who is adding exposure? Who is reducing it? Are traders opening new positions or closing existing ones? Is current positioning genuinely unusual compared with its own history?

That is where we start.

Where the Data Comes From

Our COT research is based on data published by the U.S. Commodity Futures Trading Commission (CFTC).

For physical commodity markets such as Gold, we primarily use the Disaggregated Commitments of Traders — Futures Only report. It separates reportable traders into four broad groups:

• Producer/Merchant/Processor/User
• Swap Dealers
• Managed Money
• Other Reportables

These categories are useful, but they should not be treated as perfect descriptions of every participant within them. Two traders classified as Managed Money, for example, may use very different strategies and have very different reasons for holding their positions.

The same applies to commercial participants. A producer hedging future production is not doing the same thing as a speculative trader taking a directional view, even though both may appear in the same futures market. For that reason, the category name is only the starting point.

Looking Beyond the Net Position

Net positioning is one of the most widely quoted COT figures:

Net Position = Long Positions − Short Positions

Suppose Managed Money moves from +100,000 contracts to +130,000 contracts in one week. It is clear that the group has become more net long, but the net number does not tell us how that happened.

The change could have come from 30,000 new long contracts. It could also have resulted largely from short covering. Long and short positions may even have changed at the same time. Those situations can produce similar net figures while describing very different behavior.

That is why Market Clues looks at gross longs and gross shorts alongside the net position whenever possible. We want to understand where the change came from rather than focusing only on the final number.

Observation Date and Publication Date

COT data has an important timing characteristic that is easy to overlook. The report generally reflects positions held on Tuesday, while the information is normally released to the public on Friday.

Those two dates should not be treated as interchangeable.

When researching historical data, we cannot assume that Tuesday's COT figures were already known on Tuesday if they were not released until Friday. Doing so would introduce look-ahead bias and make a historical analysis appear more useful than it could actually have been in real time.

Market Clues therefore distinguishes between the observation date, when the positions were recorded, and the publication date, when the information became publicly available.

Extreme Positioning Is Not a Reversal Signal

COT data tends to attract the most attention when positioning reaches an extreme. Managed Money may be heavily long, commercial traders may hold unusually large short positions, or a COT Index may approach the top or bottom of its historical range.

An extreme reading can be interesting, but it is not automatically a reversal signal. Strong trends can produce extreme positioning that persists for weeks or months. In some cases, an already unusual position becomes considerably more extreme before the market eventually changes direction.

When positioning reaches an unusual level, we therefore look at the details around it. How quickly did the position build? Was the move driven by fresh longs, new shorts, short covering or long liquidation? Did open interest expand at the same time? What was price doing while the positioning changed?

The extreme itself is a clue. It is not the conclusion.

The 3-Year COT Index

One way we place current positioning into historical context is with a rolling 3-Year COT Index, usually based on approximately 156 weekly observations.

The calculation is:

COT Index = ((Current Net Position − 156-Week Minimum) / (156-Week Maximum − 156-Week Minimum)) × 100

A reading near 100 means the current net position is close to the highest level observed during that three-year period. A reading near 0 means it is close to the lowest.

The important point is that the index measures relative positioning. It does not tell us that a market is overbought, oversold or about to reverse.

Its purpose is much simpler: it helps us judge whether today's positioning is ordinary or unusual compared with its own recent history.

Why Open Interest Matters

Absolute contract numbers can also be misleading when viewed over long periods. A net position of 100,000 contracts may represent a very large share of a market at one point in history and a much smaller share several years later if total futures participation has grown.

Open interest gives us additional context by showing the overall number of outstanding contracts in the market.

For some comparisons, Market Clues may therefore examine both absolute positioning and positions relative to open interest. We prefer to keep the original contract figures visible rather than replacing them entirely with a normalized measure.

Managed Money Is Not Automatically “Smart Money”

Managed Money is particularly useful when studying speculative participation because it includes professionally managed futures activity. That does not mean the category should simply be labeled “smart money.”

Professional traders can become crowded, chase established trends, reduce exposure too early or remain positioned in one direction for a long time.

The more useful questions are about behavior. Are long positions increasing? Are shorts being covered? Is exposure being reduced? Is the group already heavily committed relative to its recent history?

Those are things the data can actually help us investigate.

Commercial Positions Need Context

The Producer/Merchant/Processor/User category is also easy to misinterpret.

These participants often use futures to manage risks connected with the physical commodity. A producer may sell futures to hedge future production, while a commercial user may use futures to manage exposure to future input costs.

A commercial short position therefore does not automatically mean that the participant expects prices to fall. Commercial positioning can still reveal useful information about hedging activity and changes in market participation, but the economic reason behind these positions should not be ignored.

Nonreportable Does Not Mean Retail

Another common shortcut is to describe Nonreportable Positions as retail traders.

The COT report does not provide enough information to support that conclusion. Nonreportable positions represent the portion of open interest that is not included in the individually reportable categories. The report does not identify exactly who those traders are or why they hold their positions.

For that reason, Market Clues uses the actual term Nonreportable Positions rather than relabeling the group as retail.

What Makes a Weekly Report Interesting?

Not every weekly COT release contains a major development, and we do not try to manufacture one.

A report becomes more interesting when something meaningful changes. Managed Money may add longs aggressively after several quiet weeks. A large rise in net positioning may turn out to be driven almost entirely by short covering. Open interest may fall while price continues higher. Commercial hedging may move toward an unusual historical level.

It can also become interesting when different pieces of evidence disagree. Positioning may appear supportive while price behavior, the futures curve or another part of the market tells a different story.

Sometimes the absence of a meaningful change is useful information too.

What the COT Report Cannot Tell Us

COT data has real limitations. It is aggregated, published with a delay and divided into broad trader categories. Reporting thresholds matter, classifications can change, and the report does not reveal the exact motivation behind every position.

Most importantly, the COT report cannot tell us what price must do next.

A large long position can become larger. An extreme can persist. Commercial hedging can increase during a rising market, and speculative positioning can remain one-sided throughout a strong trend.

That is why Market Clues does not reduce COT analysis to labels such as Bullish, Bearish, Buy or Sell. The report is more useful when treated as evidence rather than an instruction.

How We Use COT Data at Market Clues

When a new report arrives, our first question is usually simple: What changed?

From there, we look at who changed their positions, whether longs or shorts were responsible, how unusual the current exposure is, whether open interest supports the change and how the latest figures compare with history.

As Market Clues adds more research layers, COT positioning can then be compared with futures structure, seasonality, volatility and related markets.

Sometimes those clues point in the same direction. Sometimes they do not. Both situations are useful. In fact, disagreement between different parts of the market can be more interesting than a perfectly consistent picture.

Data Used by Market Clues

Primary source: U.S. Commodity Futures Trading Commission (CFTC)
Primary commodity report: Disaggregated COT — Futures Only
Default COT Index lookback: approximately 156 weekly observations

Approach: Start with what changed, understand how it changed, and only then consider what it may mean.

How We Read Futures Curves

Most people first meet a futures market through a price chart. Gold is trading at one price, crude oil at another, and the chart shows how that price has moved over time. But a futures market usually has more than one price at the same time.

There may be a contract for delivery next month, another a few months later, and several more stretching further into the future. Put those contracts next to each other and you get the futures curve. That curve can reveal things the headline price does not.

Sometimes the differences between contracts barely move. At other times the front of the curve changes quickly while later contracts remain almost untouched. A market can look quiet on the outright chart while something much more interesting is happening between delivery months.

That is why we look at futures curves at Market Clues. Not because they predict the future, but because they can help us understand how pricing, availability and pressure are distributed across the market.

What a Futures Curve Actually Shows

A futures curve is simply a series of prices for different delivery dates in the same market. If nearby Gold trades at 2,400, a contract several months out at 2,415 and a later contract at 2,440, the market has an upward-sloping term structure. If nearby crude oil trades at 82 while later contracts trade at 80 and 78, the curve slopes the other way.

When later contracts trade above nearby contracts, the market is generally described as being in contango. When nearby contracts trade above later contracts, it is generally described as being in backwardation.

Those labels are useful because they give us a quick description of the curve, but they are only the beginning. Two markets can both be in contango while having very different structures. One may have only a small gap between nearby and deferred contracts, while another has a much steeper curve. The same applies to backwardation.

For research, the size and location of those differences often matter more than the label itself. We want to know which contracts are moving, how quickly the relationships are changing and whether the current structure is normal for that market. A futures curve is therefore not just a picture of prices across maturities. It is a way of looking at the relationships between those prices.

Contango, Backwardation and Carry

An upward-sloping curve is sometimes interpreted as the market expecting higher prices in the future. That is too simple.

For a storable commodity, holding the physical asset through time can involve financing, storage, insurance, handling and other expenses. These costs are part of what is commonly described as the cost of carry. If it costs money to buy a commodity today and hold it until a later delivery date, it is perfectly reasonable for a deferred futures contract to trade above the nearby market.

Gold is a good example. It is highly storable, but storing it is not economically free. Capital tied up in physical Gold has a financing cost, and storage and insurance can add further expenses. A higher deferred futures price may therefore reflect the economics of carrying Gold through time rather than a simple prediction that the spot price will rise.

There is another side to the relationship. Sometimes having the physical commodity available now has real economic value. A refinery needs crude oil to operate today, not six months from now. A manufacturer may need a particular raw material to keep production running. This benefit of immediate availability is often discussed in terms of convenience yield.

When supply is comfortable, immediate access may not be particularly valuable. When inventories tighten or a commodity becomes harder to obtain, nearby availability can become more important. Nearby contracts may then strengthen relative to deferred contracts, sometimes pushing the market toward backwardation.

This is why contango and backwardation should not be treated as simple bullish or bearish signals. The shape of the curve can reflect financing, storage, inventories, hedging, seasonality, physical constraints and expectations at the same time. The curve describes a market relationship; it does not tell us what price must do next.

Why Calendar Spreads Matter

One of the most useful ways to study the futures curve is through calendar spreads. A calendar spread compares two contracts on the same market with different delivery dates, such as June Gold versus December Gold or July crude oil versus December crude oil.

The exact calculation needs to be stated clearly because spreads can be quoted using different conventions. Once that convention is fixed, however, the spread gives us a clean way to track how one part of the curve is behaving relative to another.

Suppose crude oil rises from 80 to 82. On the outright chart, that looks like a straightforward two-dollar move. But imagine the nearby contract rises by two dollars while a contract six months further out hardly moves. Now we know something else: the front of the market has strengthened relative to the back.

That might lead us to investigate inventories, near-term supply, refinery demand, transportation constraints or another factor affecting immediate availability. The price chart tells us that oil moved. The spread tells us where within the futures structure the pressure appeared.

Calendar spreads are particularly useful because changes in the curve can begin before the overall curve changes category. A market may remain in contango while the contango steadily narrows. It may remain backwardated while the nearby premium gradually weakens. In both cases, the label has stayed the same, but the structure has changed.

What Changes in the Curve Can Reveal

A static description of the curve is useful, but changes in the curve often contain more information.

If crude oil has been in contango for several months, saying that it remains in contango each week tells us very little. If the front of the curve suddenly begins to strengthen and nearby spreads move rapidly toward zero, something has changed even though the market may still technically be in contango.

We may look at whether the curve is becoming steeper or flatter, whether nearby contracts are gaining on deferred contracts, whether a change is concentrated in one part of the curve or visible across several maturities, and how quickly those relationships are moving.

Historical context also matters. A spread may look large in absolute terms but still be completely normal for that market. Another spread may appear modest but sit near an extreme relative to its own history. Measures such as historical ranges, percentiles or standardized comparisons can therefore help answer a more useful question: How unusual is the current structure?

The direction of the outright price does not have to match what is happening inside the curve. A commodity can fall while nearby spreads strengthen, or rise while the front of the curve begins to weaken. That is not necessarily a contradiction. A broad macro move may be pushing the whole price level in one direction while conditions near the front of the market are changing in another.

Those situations are particularly interesting because they show why the curve should be studied alongside the headline price rather than treated as a secondary version of it.

Inventories, Seasonality and Physical Conditions

In many physical commodity markets, inventories and futures structure are closely related.

When stocks are abundant and storage capacity is readily available, there may be less urgency to obtain the commodity immediately. Carrying inventory into the future may be relatively easy, and contango can become more pronounced. When stocks fall, nearby supply can become more valuable. Calendar spreads may strengthen and the curve may move toward backwardation.

The relationship becomes especially useful when both datasets move together. Falling inventories combined with stronger nearby spreads can provide a more convincing picture of tightening physical conditions than either observation alone.

Inventory data still needs to be treated carefully. Published figures may cover only particular regions, storage facilities or grades. Some stocks are private, and some data arrives with a delay. The inventory measure that is easiest to obtain may not always represent the part of the physical market that matters most for a particular futures contract.

Seasonality adds another layer. Natural gas is a clear example because demand, storage and weather risks vary sharply through the year. Winter and summer contracts do not represent identical economic conditions separated only by time. Agricultural contracts can reflect planting, harvest and crop-year effects.

This means a futures spread should not always be judged against one fixed historical benchmark. A relationship that appears unusually wide in one month may be completely normal for that point in the seasonal cycle. For markets with strong seasonal structures, we therefore want to know not only whether a spread is large, but whether it is unusual for that time of year.

Contract Rolls and Data Problems

Futures-curve research becomes more complicated when we move from theory to actual historical data.

Futures contracts expire. As expiration approaches, trading activity usually shifts into a later contract. The contract we call the “front month” therefore changes repeatedly through time.

If we simply connect one front contract to the next, the resulting historical series can contain jumps that were caused by the contract roll rather than by an actual market move. An expiring contract might trade at 80 while the next active contract trades at 82. Switching from one to the other creates an apparent two-dollar increase even if neither contract actually moved.

Continuous futures series attempt to deal with this problem in different ways, often by adjusting historical prices. Those series can be useful for long-term price analysis, but they can be problematic when our objective is to study the relationship between actual contracts. For term-structure research, the individual contracts usually matter.

Liquidity matters too. Some delivery months trade actively while others may have very little activity. A settlement price in a thin contract can produce a misleading spread. Delivery rules, contract specifications and exchange changes can also affect historical comparisons.

This is why contract selection and roll methodology should never be invisible details. If Market Clues publishes a spread, curve statistic or historical comparison, we should be able to explain which contracts were used and how the series was constructed. The calculation is only as useful as the data underneath it.

Why the Curve Is Not a Forecast

One of the most common misunderstandings is to treat each futures price as a prediction of where the spot market will trade on that future date.

If December Gold trades above June Gold, that does not simply mean the market believes Gold will rise to the December futures price. If deferred crude oil trades below nearby crude, that does not simply mean traders expect spot oil to fall.

A futures price is a tradable price today for a particular delivery date. It reflects the current relationship between financing, storage, physical availability, hedging activity, arbitrage and expectations. Expectations can influence the curve, but they are only part of the picture.

This distinction is important because otherwise normal carry relationships can easily be mistaken for forecasts. A market in contango may look as though it is predicting higher prices when much of the difference simply reflects the economics of holding the asset.

At Market Clues, we therefore treat the curve primarily as a measure of market structure rather than a list of future price targets.

The same caution applies to roll yield. A futures investor who maintains exposure over time must eventually replace an expiring contract with a later one. In contango, that later contract may be more expensive; in backwardation, it may be cheaper.

Those relationships can influence long-term futures returns, but the result depends on the exact contracts used, the timing of the roll, changes in the curve and other components of return. Contango does not guarantee a loss, and backwardation does not guarantee a profit. The mechanics matter.

How We Use Futures Curves at Market Clues

When we look at a futures curve, we start with the actual structure rather than trying to force it into a bullish or bearish label. We want to know which contracts are being compared, how the curve has changed, whether nearby contracts are behaving differently from deferred ones and whether the current relationships are unusual relative to history.

Then we add context. Could seasonality explain the structure? Are inventories moving in the same direction? Is open interest changing? Are there liquidity or expiration effects that could distort the comparison? Does positioning from the COT report support the same interpretation?

Sometimes several pieces of evidence fit together neatly. At other times they do not. Suppose Managed Money has built a very large long position while the nearby futures structure is weakening. The positioning data shows strong speculative participation, while the curve suggests that the front of the market is becoming less firm.

We do not need to decide immediately that one of those observations is wrong. They may reflect different forces or different time horizons. The disagreement itself is useful.

That is one of the main ideas behind Market Clues. Positioning, futures structure, inventories, seasonality, volatility and related markets do not have to tell the same story. When they diverge, that can be just as informative as when they confirm one another.

A futures curve is therefore not another signal that we add to a score. It is another way of observing the market. For our research, we focus on actual contract relationships, clearly defined calendar spreads, changes in curve shape, historical and seasonal context, contract liquidity, expiration effects and relevant physical-market data. Whenever contract-selection or roll rules affect the result, those rules should be transparent.

The basic idea is straightforward: the headline price shows where the market is trading, while the futures curve helps us understand how the market is structured around that price.