Futures & Market Research Glossary

A seasonal window is an explicitly defined recurring time-selection rule used to isolate a period for seasonal analysis. Operationally, it specifies how the interval begins and ends and how dates are aligned—by calendar date, trading-day count, or a recurring event—so that every eligible observation can be classified reproducibly as inside or outside the window. For example, “from the first trading day of November through the fifteenth trading day of December, inclusive” is an operational window rule; “early winter” is not. The window identifies when a possible seasonal tendency is measured. It does not establish the direction or certainty of a market move.

The window rule and the study specification

The temporal rule should state the start condition, end condition, alignment convention and any boundary treatment that can change membership in the interval, such as whether endpoints are inclusive and how non-trading days are handled. Reproducing the resulting seasonal study requires additional information that is not strictly part of the window itself: the instrument or contract series, the observation or price field, the variable being measured and the historical sample. For example, two researchers can use exactly the same November–December window but obtain different return histories if one uses settlement prices from a front-contract series and the other uses a differently constructed continuous futures series.

Calendar time is not always economic time

A fixed calendar window is most defensible when the underlying mechanism is itself reasonably stable in calendar time. Some recurring influences move from year to year. Holiday dates can shift, crop-development milestones can depend on growing conditions, and operational schedules can change. In those cases, a rule defined relative to an event or trading-day count may represent the economic mechanism more faithfully than identical calendar dates every year. Weekday composition, leap years and exchange holidays can also alter which observations fall inside apparently similar calendar intervals.

Futures research adds a further implementation problem. A continuous series can move from one contract to another under different roll or contract-selection rules. If those transitions repeatedly fall inside the seasonal window, part of the measured pattern may reflect series construction rather than only the underlying seasonal force. The relevant contract and continuous-series methodology should therefore be documented and, when consequential, tested against reasonable alternatives.

Window selection also affects statistical credibility. If start and end dates are chosen only after searching many alternatives for the strongest historical result, the reported tendency is exposed to data-snooping and overfitting risk. Out-of-sample testing, stability across reasonable specifications and a continuing economic rationale provide stronger evidence than an optimized in-sample window alone. Once the rule is fixed, researchers can examine returns, spreads, volatility, inventories or other variables within it across years and regimes. Those statistics describe historical behavior conditional on the chosen window; they do not make the window a deterministic forecast or a standalone trading signal.

Seasonality is the tendency for a market variable to show a recurring pattern at roughly the same part of the calendar year because underlying economic, biological, operational or behavioral forces also recur. In commodity markets it can appear in production, demand, inventories, spreads, volatility and prices: harvests arrive on a schedule, heating and cooling demand shifts with weather, storage is built and drawn, and marketing activity changes through the crop year. The key word is tendency. Seasonality describes a historically recurring pattern, not a deterministic forecast of the next move.

That calendar link separates seasonality from broader cycles. A commodity or business cycle may expand and contract without returning at the same time each year; a seasonal pattern is anchored, however imperfectly, to recurring dates, months or parts of the year. The economic mechanism matters more than the shape of the historical average. Two markets can display similar monthly price patterns for different reasons, while the same market can lose an old seasonal pattern when production geography, imports, storage capacity, technology, regulation or demand behavior changes.

A seasonal average can outlive its mechanism

Seasonal studies therefore need more than a multi-year average. The sample period, calendar alignment, treatment of outliers and choice of mean versus median can change the apparent pattern, especially when a few unusual years dominate the history. Stability should be tested across subperiods and regimes, and the current economic mechanism should still be plausible. A pattern that was persistent before a structural change may remain visible in a long sample even after it has weakened in recent years.

In futures, construction is part of the question

Researchers must also define what is being made seasonal. Outright price levels, returns, calendar spreads, inventories and volatility measure different phenomena. Futures add another complication because a long history is usually assembled from contracts that expire and are replaced. Contract-selection and roll rules can therefore affect the timing or magnitude of an apparent seasonal pattern; an unadjusted jump created when one contract is replaced by another should not be mistaken for a recurring market move.

Seasonality is most useful as a conditional research framework: it identifies periods when recurring forces have historically mattered and focuses attention on whether those forces are present now. It becomes misleading when the historical pattern is treated as a stand-alone signal, when the sample is unstable, or when data construction is allowed to masquerade as economics.

A settlement price is the official daily price established for an exchange-traded futures contract and used to mark open positions, calculate daily gains and losses, and support margin and clearing processes. It is often described as the market’s daily close, but that shorthand can mislead: the settlement price is determined under an exchange’s stated methodology and therefore need not equal the last transaction price of the session. In this entry, the term is used in its futures-market sense across commodity and financial futures, including physically delivered and cash-settled contracts; settlement conventions for options and bilateral OTC derivatives are outside its scope.

Why settlement can differ from the last trade

For many contracts, the daily settlement price is determined from activity during a defined settlement window. Depending on the product and its liquidity, the methodology may use qualifying trades, a volume-weighted average price, bids and offers, spread relationships between contract months, or other prescribed market inputs. If a contract is thinly traded, the exchange may rely on related contracts, spreads, quotes, or other fallback procedures rather than copy a single late trade. Settlement is therefore a rule-based clearing reference, not simply the final print on the screen.

The timing creates an important research trap. For some contracts, the settlement window ends before trading ends for the session. A data vendor’s end-of-session “close” may therefore incorporate trades that occurred after the information window used to determine the official settlement. Substituting close for settle—or settle for close—can change daily returns, marked profit and loss, volatility estimates, rolling signals, and continuous-futures construction. Researchers should identify which field their data set actually contains and align it with the applicable contract’s settlement procedure and timestamp.

Daily settlement and final settlement are not the same concept

Daily settlement is the recurring mark used through the life of the contract. At expiration, a contract may instead be assigned a final settlement price under a separate rule. Cash-settled futures can reference an index fixing, special quotation, average benchmark rate, or another contract-specific calculation; physically settled futures can use settlement values in delivery or invoice mechanics according to their rules. The last daily settlement and the final settlement may coincide for some products, but this should not be assumed. When expiry, delivery, backtesting, or historical profit and loss matters, the contract’s settlement methodology—not a generic “closing price” field—is the authoritative reference.

In commodity and futures analysis, the spot price is the current cash-market price for the underlying commodity for immediate or customary near-term delivery. It refers to the actual or cash commodity at a specified time and place, so it is distinct from the price of a futures contract for delivery in a later contract month.

Spot is a delivery convention, not a contract month

The word “immediate” is less literal than it sounds. What qualifies as spot delivery follows the commercial conventions of the market: the accepted delivery window, location, grade or quality, and other transaction specifications can differ by commodity. Official energy-market definitions, for example, describe spot transactions as immediate or near-term delivery rather than imposing one universal same-day rule. A spot quote therefore has meaning only with its underlying specification; two prices labeled “spot” need not represent the same deliverable economic exposure.

This distinction becomes especially important beside futures. The front-month contract is still a futures price, even when expiration is close. The difference between a comparable cash price and futures price is the basis. As a physically delivered futures contract approaches delivery, the delivery mechanism normally encourages cash and futures prices to converge, but comparison requires like-for-like location, quality, and timing. A persistent gap may reflect those specification differences rather than a simple pricing error.

What researchers should verify before using a spot series

The term also appears in financial markets, but it does not identify the same kind of object in every market. In foreign exchange, a spot rate is the exchange rate agreed for an exchange of two currencies with cash settlement within two business days. For equity-index futures, “spot” or “cash” commonly refers to the current value of the underlying cash index—a calculated benchmark rather than a directly deliverable asset. Both are cash-market references, but one is a transaction rate and the other is an index level.

That distinction matters for data construction. In physical markets, a useful spot series should identify what is being priced, where, and under which delivery convention. In financial markets, the researcher should establish whether “spot” refers to a tradable cash transaction, a calculated reference level, or another market-specific cash benchmark. Substituting a nearby futures settlement for spot, or combining series with incompatible specifications, can distort basis, carry, hedge effectiveness, and convergence analysis. The practical rule is simple: spot is the current cash-market reference for a defined exposure, not merely the nearest price on the screen.

Spreading Positions are the portion of a reportable trader’s Commitments of Traders (COT) position that the CFTC computes from qualifying offsetting long and short exposure within the same reported market. They are not a separate trader classification and should not be read as a complete measure of every spread strategy a trader may hold. The key distinction is that “spreading” is a reporting decomposition: it identifies the matched part of a trader’s long and short position, while any unmatched remainder stays in the long or short column.

How the matched amount is constructed

In a futures-only report, qualifying spreading reflects offsetting futures positions in different calendar months. In a Futures and Options Combined report, futures and options can contribute to the matched amount in the same or different calendar months, with option positions incorporated on the CFTC’s futures-equivalent, delta-adjusted basis. Inter-market spreads are excluded. If a trader has 120 qualifying long contracts and 80 qualifying short contracts, for example, 80 may be reported as spreading and the remaining 40 as long. Because the calculation is made at the trader level before positions are aggregated into public categories, the published spreading figure is not obtained by simply netting a category’s aggregate long and short totals.

The column depends on the COT report family

Spreading is not reported identically across COT classifications. In the Legacy COT, the spreading column applies to Non-Commercial positions; Commercial positions are shown as long and short. The Supplemental COT likewise shows spreading for Non-Commercial positions, while Commercial and Index Trader positions are reported as long and short. In the Disaggregated COT, spreading is reported for Swap Dealer, Managed Money and Other Reportables, but Producer/Merchant/Processor/User positions are reported only as long or short. In the Traders in Financial Futures (TFF) report, spreading is reported for all four categories: Dealer/Intermediary, Asset Manager/Institutional, Leveraged Funds and Other Reportables.

This category structure matters because a large spreading number does not by itself mean that the trader group is economically neutral, pursuing a calendar-spread strategy, or expressing no directional view. A trader can have a matched spreading component and a residual outright position at the same time; accordingly, the same trader may also appear in long or short trader counts. The CFTC notes that this partial-activity treatment can make the sum of category trader counts exceed the total number of reportable traders.

For research, one subtle consequence is that spreading can obstruct simple reconstruction across report formats. The CFTC states that subtracting Futures Only data from Futures and Options Combined data can recover much of the options information, but some information is lost because of spreading. Treat the published spreading column as a specific CFTC aggregation rule, not as a universal label for spread trading or a directional signal.

Standard deviation is a measure of how widely observations are dispersed around their mean. It is the square root of variance, so it is expressed in the same units as the data rather than in squared units. A larger standard deviation means the observations are more spread out; a smaller value means they cluster more tightly around the mean.

The denominator changes the estimate

For a complete population, standard deviation uses the average squared deviation from the population mean and then takes the square root. When a sample is used to estimate population variability, a common sample formula divides the sum of squared deviations by n − 1 before taking the square root: s = √[Σ(xᵢ − x̄)²/(n − 1)]. The denominator is not a cosmetic choice. Using n instead of n − 1 produces a different value, with the difference most noticeable in short samples. For an independent random sample from a population with finite variance, the n − 1 correction makes the sample variance unbiased; taking the square root does not generally produce an unbiased estimator of the population standard deviation.

When dispersion is used as volatility

In market research, standard deviation is often applied to returns to quantify realized dispersion over a chosen window. That number is inseparable from its construction: daily versus weekly observations, arithmetic versus logarithmic returns, lookback length, missing-data treatment and the denominator convention can all change the result. Standard deviation calculated on price levels answers a different question; it measures dispersion of those price observations around their mean and should not automatically be interpreted as return volatility.

Because deviations are squared, extreme observations receive disproportionate weight. That makes standard deviation responsive to large market moves but also sensitive to outliers and heavy tails. It does not imply that the underlying distribution is normal, nor does one standard deviation carry a fixed probability meaning without a distributional assumption. A volatility regime shift can also make a single full-sample value conceal important time variation.

For reproducible market research, the important point is therefore not merely the reported standard deviation but what was measured, over which window, at what frequency and with which denominator. Those choices should be aligned before values are compared across markets, models or software implementations.

The stocks-to-use ratio measures end-of-period inventories relative to a defined measure of use over an accounting or marketing period. It is normally expressed as a percentage: ending stocks divided by the applicable use measure, multiplied by 100. A 15% ratio therefore means that ending stocks equal 15% of the denominator used in that particular balance sheet. Scaling stocks by use makes the ratio more informative than the inventory level alone: the same physical quantity of stocks can represent a large cushion in one market and a much thinner cushion in another.

The denominator is part of the measurement

There is no completely universal construction of “use.” Official series may employ different balance-sheet definitions depending on the commodity, geographic scope and dataset. USDA cotton documentation, for example, calculates stocks-to-use from ending stocks divided by total disappearance, with total disappearance defined as mill use plus exports. A USDA rice table instead calculates the ratio from ending stocks divided by its defined total domestic-use measure. Researchers should therefore verify the denominator, geographic coverage, marketing-period convention and measurement basis before comparing ratios across commodities, countries, datasets or historical series. Two ratios carrying the same label are not necessarily constructed identically.

A smaller buffer is not a price forecast

A lower stocks-to-use ratio generally describes a smaller inventory cushion relative to use and can be associated with tighter physical-market conditions; a higher ratio describes a larger cushion. The percentage does not, however, reveal whether all reported stocks are equally accessible to the market. Location, ownership, quality, policy constraints and other limits on availability can change the economic meaning of the headline inventory. USDA research has noted, for example, that an identical quantity of privately held stocks and government-controlled stocks may be perceived differently because their release into the market can operate differently. The ratio is therefore most useful as a consistently defined measure of relative inventory pressure, not as a mechanical bullish or bearish signal. Its interpretation is strongest when the researcher also understands what is counted as stocks, what is counted as use, and how readily those stocks can actually satisfy demand.

A storage cycle is the recurring process by which a storable commodity is accumulated during periods when supply is relatively abundant and released when the value of making that supply available now exceeds the value of retaining it for later. It links production, consumption, trade, logistics and storage across time. Storage cycles are often seasonal, but the calendar does not itself trigger accumulation or withdrawal; changing physical balances and intertemporal economics do.

Storage transfers physical supply through time

Many commodity markets do not produce and consume at the same rate throughout the year. Surplus supply can therefore be carried forward rather than consumed immediately. Agricultural inventories, for example, can build after a concentrated harvest and decline as the stored crop is subsequently consumed, processed or exported. Natural-gas inventories commonly increase during lower-demand periods and decline during higher-demand periods, although weather, production and regional conditions can materially alter that pattern.

The release trigger is economic as well as physical. Inventory tends to be drawn when immediate consumption cannot be met as economically from current production or other inflows, when a disruption increases the value of readily available supply, or when selling or using the commodity now is more attractive than continuing to hold it after expected future value and carrying costs are considered. Conversely, inventory tends to accumulate when current supply is relatively plentiful and retaining the commodity for future use has sufficient value to justify storage.

In research terms, a storage cycle emphasizes this process of transferring physical supply between periods, whereas an inventory cycle emphasizes the resulting evolution of measured stock levels. The distinction matters because an inventory number alone does not reveal why stocks are being held or released, nor every constraint governing how much commodity can be stored, where it is located or how quickly it can reach the market.

Inventory volume is not the same as deliverability

Underground natural gas storage makes this distinction especially visible. Base or cushion gas is maintained primarily to support reservoir pressure and operation, while working gas is the portion available to the market. Even the amount of working gas does not fully describe short-term supply flexibility: withdrawal capability is variable and generally declines as gas is removed from a reservoir, while injection capability also changes with the amount already in storage. Facility type, equipment and operating constraints further affect how rapidly inventory can be cycled.

For market research, a storage cycle is therefore most informative when inventory levels are considered alongside production and consumption balances, available capacity, location, injection and withdrawal capability, logistics and carrying economics. A storage build or draw records the outcome of those changing incentives and constraints; by itself, it does not establish future price direction.

A structural break is a discrete change in one or more parameters or relationships in the process represented by a statistical model. The change may affect a mean, trend, regression coefficient, volatility parameter or another modeled relationship, so that one fixed set of parameters no longer describes the entire sample adequately. The defining feature is therefore not simply a large movement in the observed data, but evidence that the relationship generating or governing those observations has changed.

A break changes the relationship, not merely the observation

An isolated shock can create an extreme observation without producing a structural break if the previous statistical relationship subsequently resumes. A structural break instead implies parameter instability across different parts of the sample. It may affect all model parameters or only some of them, and a dataset can contain one break or several. This also distinguishes a structural break from a data revision: a revision changes previously recorded or published observations because the underlying data or methodology has been updated, whereas a structural break concerns a change in the behavior of the process or relationship being modeled. Historical observations can remain exactly as recorded while the model linking them becomes unstable.

How the possible break date is determined matters for inference. If a candidate date is specified independently of the sample being tested, statistical inference can condition on that proposed point. If the break date is unknown and researchers search across many candidate dates, however, they cannot generally select the most dramatic split and then treat it as though it had been specified in advance. Structural-change procedures designed for unknown break points account for that search, while multiple-break methods address situations in which parameters change more than once. An estimated break date should also be understood as a statistical estimate, not necessarily the exact economic moment at which a new regime began.

Why ignoring a break can change the conclusion

Structural breaks matter because many statistical methods rely, explicitly or implicitly, on parameter stability over the estimation window. Pooling observations from different regimes can therefore produce coefficients, correlations, forecasts or long-run relationships that describe neither regime particularly well. One important time-series consequence is that an unmodeled shift in level or trend can make an otherwise stationary process appear unusually persistent or even consistent with a unit root. Structural change can consequently alter unit-root and cointegration inference rather than merely create an obvious discontinuity in a chart.

Detecting a structural break does not by itself identify its economic cause or prove that the latest regime will continue. Policy changes, market structure, liquidity conditions, technology or major shocks may coincide with parameter instability, but establishing causation requires separate evidence. For market research, the practical question is therefore not simply whether the latest observation looks unusual, but whether the historical relationship itself remains stable enough to justify estimating and interpreting it as one process.

The Supplemental COT Report is a weekly Commodity Futures Trading Commission (CFTC) report family that isolates Commodity Index Trader positioning within selected agricultural markets. The current Supplemental report covers 13 selected agricultural commodity contracts and is published only in a Futures-and-Options-Combined format. It breaks reportable open interest into Non-Commercial, Commercial, and Index Trader classifications, with nonreportable positions shown separately. In the traditional report table, spreading is reported separately for Non-Commercial traders, while Commercial and Index Trader positions are shown as long and short.

The report normally describes positions as of the preceding Tuesday and is generally released Friday at 3:30 p.m. Eastern Time; holidays or other disruptions can alter that schedule. Because the Supplemental report is combined, option open interest and traders' option positions are converted to futures-equivalent positions using delta factors and then combined with futures positions. The resulting figures therefore should not be interpreted as raw option contract counts.

Why Index Trader is not a direct measure of index trading

The report's defining feature is the Index Trader classification, but that category is not a transaction-by-transaction identification of commodity-index activity. For Supplemental-report purposes, the CFTC applies a trader classification rather than labeling each individual position by motive. If index-related trading predominates for a trader under the Supplemental methodology, the trader's positions are classified as index related even when some positions reflect other activity.

This creates an important measurement boundary. Published Index Trader positions can overstate actual index trading when a classified trader also conducts non-index activity, and they can understate it when index exposure is internally netted against other positions before reaching the futures market. Index activity can also be absent from the category when it is not a substantial part of a trader's overall activity. The series is therefore a trader-classification aggregate, not a clean decomposition of every position by economic purpose or a direct measure of investment flows.

Relationship to other COT reports

The CFTC began publishing the COT-Supplemental in January 2007. The initial report covered 12 selected agricultural commodities, whereas the current report covers 13, so historical research should not assume that today's coverage applied from inception. This is especially relevant when constructing long time series or comparing category totals across different periods.

Index Trader also should not be treated as equivalent to the Swap Dealer category in the Disaggregated COT Report. The CFTC has noted that the Disaggregated Swap Dealer category can include swap dealers with no commodity-index business, while the Supplemental Index Trader category can include pension or other investment funds that obtain index exposure directly rather than through a swap dealer. The Supplemental report is therefore most useful when the research question concerns the CFTC's classified index-trader positioning in its covered agricultural markets, not when a position-by-position estimate of commodity-index investment, a futures-only view, or the Disaggregated report's participant taxonomy is required.