The Market Does Not Know Your P&L

Scope: definitions, arithmetic examples and a measurement design. No market data and no behavioral dataset is presented; the note says so wherever a measured version would go.

The market is unaffected by your running profit — but you are not, and that asymmetry is measurable. Two situations traders treat as separate problems, the prop-firm evaluation and the given-back winning day, are the same mechanism wearing different clothes: an account state (distance to a target, distance to a limit, size of an unrealized cushion) changing the trader's next decision while changing nothing about the market. This note defines how to measure that mechanism instead of calling it "psychology".

Why is an evaluation harder than the same strategy in a normal account?#

Because it adds a second question. A normal account asks only "does this trade have positive expectancy?" — a question you can answer correctly and still lose the individual trade. An evaluation adds "can I reach the target before violating the rules?", which is a different optimization problem: the drawdown limit, the daily loss cap and the deadline make the path matter, not just the destination.

Arithmetic makes the path problem concrete. Two accounts both finish +8R:

Identical strategy, identical final result — but under a typical evaluation drawdown rule, Account B fails at trade two and never gets to produce the profit it was always going to make. A pass/fail verdict on B measured the ordering of its variance, not its edge. That is path dependence, and it is a property of the rules, not the trader.

If the rules don't touch the market, what do they touch?#

The trader's behavior — in two opposite directions, often in the same account within the same week. Near the drawdown limit, the instinct "I cannot lose again" produces under-trading: skipped valid setups, stops moved, winners cut early, size reduced erratically. Behind the target with a deadline approaching, "I need $2,000" produces over-trading: more trades, larger size, lowered standards, marginal hours. Same trader, same strategy, opposite distortions — selected by account state.

None of this needs to be discussed as character. Every item in it is observable in a trade log: frequency, size, time-between-trades, stop distance, setup quality before versus after crossing 25% / 50% / 75% / 90% of the target. If those distributions shift as the state changes, the evaluation is altering the trading it claims to be measuring.

What does a winning day have to do with any of this?#

It is the same state variable with the sign flipped. A trader up $2,000 by mid-morning has a market unchanged from 9:30 — but their reference point has moved from "is there a setup?" to "can I make it $3,000?", and the behavioral menu that follows (more trades, bigger size, thinner setups, less patience) is the target response from the evaluation, running without any prop firm in sight.

The useful metric here is the giveback ratio: (peak session P&L − final P&L) ÷ peak session P&L. A day that peaks at +$2,000 and closes at +$800 is not a losing day — but it gave back 60% of a result already achieved, and the final number alone hides that entirely. Two traders finishing +$500 are indistinguishable in a journal that records only outcomes; the one who was up +$2,500 first traded a very different afternoon.

What would this look like measured properly?#

Bucket every session by P&L state — below −1R, near flat, +1R to +2R, above +2R — and compare the next trade across buckets: size, frequency, win rate, average R, adverse excursion. The question is not whether traders feel different at +2R; it is whether the expectancy of the marginal trade changes with the state. If expectancy holds up after +2R, continuing is right and "quit while ahead" is superstition. If it deteriorates sharply, the data has located a stopping threshold that no influencer rule could. The same design runs on the loss side: what actually happens after −2R — longer waits, or faster and bigger?

This note publishes the design and the definitions; the bucketed distributions are not yet published. What requires no data is the conclusion of the arithmetic above: any rule or reference point that makes the path matter will select for behavior the unconstrained strategy never exhibits — and a trader who has never logged their own state-dependence has never seen their real strategy, only its good moods.

What changes tomorrow#

Add one column to the journal: session P&L at the moment of entry. Nothing else changes. After a few weeks that single column answers, with your own data, whether your +2R self and your −2R self trade the same system.

Related reading#

• Why a winning strategy can still blow up the account — the sizing arithmetic that decides whether the account survives its own variance.
• Before you believe a reversal story — the same insistence on definitions and controls, applied to what the chart seems to show.