Behavioral Pattern Claims: What Makes One Trustworthy?
A behavioral pattern claim needs a binary trigger, a 20-trade per-pattern sample, a stated confidence interval, and per-setup separation before it surfaces.
By Imperial Analytics
A behavioral pattern claim is one of the more useful outputs a trade journal can produce, and one of the easier ones to get wrong. The right version of the claim names the conduct, the sample, and the uncertainty around the figure. The wrong version names a dollar number and lets the reader fill in the rest. This primer sets the conditions a behavioral pattern claim has to meet before it earns a surfaced spot in a journal report, and shows how each condition shows up at the row level.
What a behavioral pattern claim actually is
A behavioral pattern claim states that a defined trader conduct, such as a revenge entry inside eight minutes of a stop, carries a specific result, expressed as a dollar cost, an R cost, or an expectancy gap. Every claim has three required parts: the trigger definition, the result statistic, and the sample count that produced it.
A behavioral pattern claim has the shape "conduct X, when it occurs, is associated with result Y across N trades." Conduct X has to be a thing the journal can detect from columns it already records. Result Y has to be a number with units that the reader recognizes, not a vague direction. N is the count of trades that triggered the pattern, not the total number of trades in the journal.
Several conduct patterns are common enough to ship as standing claims. The revenge entry inside a short window after a stop. The stop widened mid-trade. The position taken outside the plan window. The exit at a price worse than the planned stop. Each one has a journal-detectable trigger and a result statistic the trader cares about.
What the claim is not, in the Imperial sense, is a story about a trader's mood. Moods are real, and they are part of the cause, but they are not the claim. The claim is the observable record of the conduct and its outcome. If the conduct cannot be read off the journal, the claim has nothing to stand on.
Why a behavioral pattern claim needs a precise trigger definition
Without a binary trigger, you cannot count instances, and without a count, the result statistic floats. A precise trigger defines the conduct in journal terms: a timestamp gap, a stop-distance ratio, a planned exit flag. A second reader of the same journal should classify each trade the same way the first did.
A trigger has to answer a yes-or-no question about each trade. "Was this entry placed inside eight minutes of the prior stop?" is a yes-or-no question. "Did the trader feel angry?" is not, because the journal does not record it. The good trigger reduces to a comparison between two columns that already exist, or two columns the trader is willing to add for the rest of the journal's life.
A loose trigger lets the same trade be in or out of the pattern depending on who is reading. The result statistic then changes with the classifier rather than the conduct, and the claim loses the property a claim is supposed to have, which is that two careful readers reach the same number from the same data.
The trigger also has to be set once and held. A trigger that gets retuned every time the result statistic disappoints is not a trigger; it is a search for a number. If a trigger needs to change because the definition was wrong at the start, the prior figures are retired with it, not carried forward as if the definition were stable.
The per-pattern sample-size floor that earns a surfaced claim
Imperial Analytics requires twenty trades in the matching condition before a behavioral pattern claim is surfaced. Below twenty, the standard error around the rate or cost is wide enough to swallow most plausible effect sizes. The floor is per pattern, not per account; one hundred total trades earn no claim if only twelve of them match.
Twenty trades is not a magic threshold. It is the point at which the standard error around a rate near one half falls below the half-width that would let a reasonable trader act on the figure. With ten trades the half-width sits near thirty-one percentage points; with twenty trades it sits near twenty-two; with fifty trades it sits near fourteen; with one hundred trades it sits near ten. The full math is in the sample-size primer; the floor lives one level above the math, as a discipline gate.
Data note
The twenty-trade floor is a per-pattern minimum, not a per-account minimum or a per-day minimum. The interval widths quoted above are calculated from the standard binomial standard error at a rate near one half, with the normal-approximation multiplier 1.96 for a ninety-five percent two-sided interval. These figures are illustrative arithmetic, shown to teach the method, not the trader's own counts.
The floor is also a conversation gate. A claim with eighteen matching trades is a candidate, not a finding. Imperial Analytics shows the candidate in a separate column with its current count and the count remaining, so the trader sees the claim forming rather than being told it does not exist. A trader who wants to know whether the eighteen-trade reading is going to firm up reads the column and decides whether to keep logging.
How to express the uncertainty around a behavioral claim
A behavioral claim reports the rate or per-trigger cost together with its ninety-five percent interval, not the point statistic alone. A figure that reads "your revenge-trade rate is sixty percent" is half a claim. The form that earns its place reads "sixty percent, CI thirty-eight to eighty-two, across twenty matching trades."
The reason the interval matters is that a point figure invites the trader to act as if the rate is exact. Acting on an exact rate makes the trader change behavior in the direction of the figure, and the figure was a noisy estimate of the rate, not the rate itself. The interval reminds the reader of the noise and contains it.
The interval should be width-honest, not width-friendly. At twenty trades the half-width is around twenty-two percentage points, and that is the figure the report carries. Shrinking the interval visually by hiding the lower edge, or quoting only the half that points the trader's way, defeats the discipline the interval is there for.
The same rule covers the per-trigger dollar cost. A behavioral claim that revenge entries cost the trader two hundred dollars per trigger on average should carry the standard error of that mean and the interval. Without the interval, the trader has no way to read whether the figure is a one-instance outlier or a stable per-trigger result.
Why per-setup and per-account separation matter for behavioral claims
Behavioral patterns express themselves differently inside different setups and inside different accounts. A revenge claim averaged across a momentum setup and a mean-reversion setup hides the case where one of the two carries the entire cost. Per-account separation handles the case where the trader runs a funded account at one size and a personal account at another.
Pooling across setups is the more common error of the two. A trader who runs a momentum setup and a mean-reversion setup will often find that revenge conduct shows up almost only after stops on the mean-reversion setup, because the mean-reversion setup is the one that produces tight stops in choppy conditions. The pooled rate hides the asymmetry, and the trader works on revenge across both setups when the cost lives inside one of them.
Pooling across accounts hides a different problem. A funded account run at one contract size and a personal account run at three contracts produce different per-trigger dollar costs from the same conduct. The pooled per-trigger figure averages the two sizes and tells the trader nothing about either. The split sets the per-account figure against the per-account size, which is the form the trader can act on.
The separation is not a presentation choice. It is the structure that lets the claim be true, because the conduct itself is not the same across setups and the cost is not the same across account sizes.
How to label illustrative versus real figures in a behavioral claim
Every figure in a behavioral pattern claim is either drawn from the trader's own fills or is illustrative arithmetic shown to teach the method. The two cases need different labels. Real figures travel with the trigger, the sample count, and the CI; illustrative figures sit inside a data note that names them as illustrative.
Real figures inherit their authority from the count of fills behind them. An honest claim never quotes a real per-trigger dollar cost without quoting the count and the interval next to it, and never moves an illustrative figure into the section that reports real figures without re-labeling.
Illustrative figures are not a problem; they are a teaching surface. The problem is the unlabelled mix. A paragraph that reads "your revenge entries cost two hundred dollars per trigger" without a label tells the trader the figure is from their fills, even if the figure was a worked example. Once the label drifts, the discipline drifts with it.
↳ Note
Twenty trades is not a magic threshold. It is the point at which the interval around the rate stops being wider than the claim itself.
The label rule is also the rule that makes the claim portable. A behavioral pattern claim that gets pasted into a coaching note or into a quarterly review carries its label with it, so the reader two months later does not have to remember which figures were the trader's own and which were the teaching example. The label is part of the claim, not a footnote on it.
Frequently asked questions
- q: What counts as a behavioral pattern in a trading journal? a: Defined trader conduct that is detectable in the journal record, such as a revenge entry inside eight minutes of a stop, a stop widened mid-trade, an oversized position relative to plan, or an exit at a price worse than the planned stop. Each one needs a binary trigger to be countable.
- q: Why does Imperial Analytics require twenty trades in the matching condition? a: Below twenty trades, the standard error around the observed rate or per-trigger cost is wide enough to swallow most plausible effect sizes, and the resulting interval is wider than the claim itself. Twenty is the point where the interval starts to inform a decision rather than overwhelm it. The floor is per pattern, not per account or per day.
- q: Does a behavioral pattern claim need a statistical test? a: For most journal use, a ninety-five percent confidence interval next to the rate or the per-trigger dollar cost is what the reader needs. A formal hypothesis test against a null is useful when comparing two setups or two accounts side by side, and it is reported alongside the interval, not in place of it.
- q: How should illustrative figures be labelled inside a behavioral claim? a: Inside a data note that names them as illustrative, on the same page as the figure. Real figures and illustrative figures never share a paragraph without a separator and a label. A label that lives two paragraphs away from the figure has already failed; the reader has moved on.
- q: What if a trader has fewer than twenty matching trades for a pattern? a: The claim is carried as a candidate rather than a surfaced finding. The journal report shows the current count and the count remaining to cross the floor, so the trader sees the claim forming rather than being told it does not yet exist. When the count crosses twenty, the row promotes itself to a surfaced claim with its rate, its interval, and its sample.