Calculation methods
Read the calculation, eligibility, comparison-window, and ranking explanations behind the tools.
Calculation methods and coverage
Each measurement compares one trader group with its own history. The tables below explain the calculation windows, eligibility requirements, and label boundaries.
Labels use the underlying measurements and the existing eligibility checks, not rounded display text. The levels are descriptive categories, not calibrated probabilities, expected returns, or trading recommendations.
Historical-position levels
The main range uses up to the latest 156 reports and includes the current position. Historical-range positioning signals require at least 104 valid positions. The actual count and comparison dates are shown with the reading.
| Level | When this label applies |
|---|---|
| Near historical high or low | The existing qualifying upper- or lower-range condition applies. The upper extreme zone begins at COT Index 90; the lower extreme zone ends at 10. |
| Very near historical high or low | An eligible upper reading has an unrounded COT Index of at least 95, or an eligible lower reading has an unrounded index of at most 5, unless it meets the actual boundary condition below. |
| At historical high or low | The current net position equals the actual maximum or minimum of its nonconstant comparison range. A rounded display value of 100 or 0 is not sufficient. |
At the boundary takes precedence over Very near, which takes precedence over Near. Missing or invalid range comparisons do not receive an extreme label. A position's index can change when an old high or low leaves the rolling window, even if the current net position is unchanged.
New historical high/low events remain separate from these degrees. Additional 52-report events add context but do not create a positioning signal by themselves.
Unusual weekly changes
Weekly-change comparisons use valid seven-day intervals within the last 156 report intervals and require at least 104 comparable changes. Gaps are excluded from this comparison. A displayed nonweekly difference keeps its actual dates and limitation rather than receiving an eligible weekly-change grade.
The tie-aware percentile ranks signed changes and includes the current valid interval. Tied values share their midpoint rank. If every comparable change is identical, the event percentile is 50.
p is the tie-aware percentile of a signed weekly change.
| Degree | Net increase: p | Net decrease: p |
|---|---|---|
| Large | 95 ≤ p < 97.5 | 2.5 < p ≤ 5 |
| Very large | 97.5 ≤ p < 99 | 1 < p ≤ 2.5 |
| Exceptional | p ≥ 99 | p ≤ 1 |
These subdivisions do not replace the existing sign, interval, history, and other eligibility requirements. The at-or-below percentile and the tie-aware percentile are different measures; the unusual-change grades use the tie-aware measure.
Additional measurements and ranking
A participant's Details in COT Analysis contains its available short-window indices, change percentiles, comparable-change counts, net-position z-score, and internal ranking score. Each value belongs to its labeled group and measurement window. The count of valid positions is not the count of valid weekly changes.
The internal ranking score helps order the displayed readings. It is not a signal-strength grade, probability, or trading recommendation. Positioning signals are considered first, then participant ranking. The score combines positioning extremes, changes, and net flips, and can exceed 100. Additional trader groups or repeated rule matches do not add points. Speculative groups may appear first in comparisons for convenience, without extra ranking points.
Comparing trader groups
Compare groups within the same contract, report family and date. Their different roles provide context; overlapping reports are not independent votes on a forecast.
| Report | Groups used in the tools |
|---|---|
| Legacy | Noncommercial traders and Commercial hedgers. |
| Disaggregated | Managed Money, Producer/Merchant, and Swap Dealer. |
| Financial/TFF | Leveraged Funds, Asset Manager, and Dealer. |
What the group names mean
Legacy Commercials typically use futures for business hedging. Noncommercials are large reportable traders not classified as commercial hedgers.
Disaggregated Managed Money includes commodity trading advisers, commodity pool operators, and other money managers. Producer/Merchant covers physical commodity businesses; Swap Dealers manage swap-related exposure.
Financial/TFF Leveraged Funds includes hedge funds and similar leveraged trading entities. Asset Managers include institutional investors such as pension and mutual funds. Dealers are sell-side participants such as banks and securities dealers.
These groups have different roles. All reports keeps their comparisons separate because classifications overlap. Public reports include additional trader categories beyond the subset displayed here.
Compare each group with its own history. Opposite historical extremes describe different locations within those separate ranges, not necessarily opposite current net directions. Across the complete futures market, every contract has a long and a short side; the displayed subset does not have to balance. Opposite extremes and overlapping reports are not independent confirmations of a price forecast.
Why can featured groups differ?
Positioning at a glance selects a qualifying reading separately for each signal type. Different columns can therefore describe different trader groups or reports. They are not added together into a market-wide strength score.
What this means explains the selected reading. What to watch suggests a further check using price, later reports, or other groups in the same report. It is not an entry or exit instruction.
The report's historical-pattern label concerns historical-range positions. For example, no historical-range extremes does not rule out an unusual weekly change or a net flip.
On COT Analysis, the trader group in focus is selected using the page's ranking policy. Current observations take precedence over stale ones, then qualifying positioning signals, then the internal ranking score; ties use the stable observation key. The score combines historical extremes, unusual changes and flips, with coverage penalties. It is not simply the largest weekly change or a measure of signal strength. It can differ from the groups featured in Positioning at a glance. The comparison rows show the other trader groups in each selected report.
When shown beside a new high or low, the history button explains whether the group was already near that end of its range in the preceding report. It does not identify when the condition first began. The same context is available in that group’s Details.
What does “Rule matched” mean?
Rule matched identifies a reading that meets a predefined rule. The match belongs to that report and trader group, not automatically to the group featured in the summary.
On COT Analysis, rule matches, tested conditions, and supplied values are in the relevant trader group’s Details. On Signals, inspect the matching evidence in the result.
Predefined rule matches, custom numeric screens, and general positioning-signal criteria serve different purposes and need not return the same groups. A rule match is not a separate trading recommendation.
