Stress, Load and the Pre-Tilt State

What the evidence actually says, and what it does not

The popular version of this lesson says willpower is a fuel tank, that decisions drain it, and that the correct response is to economise. That story rests on “ego depletion”, a large body of studies showing that a task requiring self-control degrades performance on the next one. When independent laboratories ran the effect together with pre-registered protocols, the average effect came out near zero, which means the strong version of the claim does not survive the evidence and should not be the basis for a rule. One part of the finding did hold up, and it is stranger than the original: what people believe about willpower changes how they behave. Participants who had been led to believe willpower is a limited resource showed the depletion effect; those told it is not, largely did not. What survives replication is a different mechanism, and it is more useful. Acute stress — the kind with a deadline attached, or a live loss on screen — shifts the balance of control from goal-directed, deliberate behaviour toward habit. Experiments that stress participants before a choice task find the behaviour becomes more habitual and less sensitive to the value of the outcome; the same shift appears in animal work and has a plausible route through cortisol and the prefrontal cortex. The practical implication is precise: under stress the plan does not win because the trader tries harder, it wins because the plan has become the habit. A written system that is practised is available in a hot state; a written system that is read for the first time in a hot state is not. Sleep is the other variable with a solid evidentiary basis, and it is badly underrated as a trading input. Sleep deprivation reliably degrades prefrontal function, raises risk-taking, and increases the weight placed on the possibility of gains while reducing sensitivity to the possibility of losses. A trader who would not size up on a boom-and-bust rumour will do something like it after five hours of sleep, and will not attribute the change to the sleep. Time pressure does something related: it narrows attention, reduces the number of factors considered, and increases the use of familiar responses — which is fine when the familiar response is the plan and expensive when it is not. The pre-tilt state, and what it is made of — Sleep debt — three nights of five to six hours: Prefrontal control down, risk appetite up, loss sensitivity down · Acute stress — a deadline, a live drawdown, a public call: Control shifts from deliberate to habitual: the habit decides ← · Time pressure — the price is moving while you decide: Attention narrows; familiar responses dominate · Real-time P&L monitoring: Physiological arousal rises and decision quality falls — the screen is part of the state ← The list is not a mood description. Every row is a measured input to decision quality, and four of them can be changed with a rule rather than with resolve.

A rule for each marker, written while the state is absent

The reason the pre-tilt state is worth studying is that it is visible before it fires. Restlessness, a rising urge to act, breathing and heart rate, the compulsion to check the account, and the sudden conviction that the next trade is the one — all of them arrive before the order does, and the honest reading is that they are readings of the trader rather than of the market. Monitoring the profit and loss in real time is part of the state rather than a way out of it: research measuring traders’ physiology found that the intensity of their reactions to moment-to-moment gains and losses was associated with worse performance, which is the opposite of what a screen full of numbers offers to do for you. Because the state removes the ability to decide well, the response cannot be a decision taken inside it. Each marker gets a rule written in advance, so the hot state only has to recognise a trigger rather than reach a conclusion. Three losses in a session halve the unit until two planned trades have been followed, as the revenge lesson has it. A daily loss limit ends the session, and closing the platform is the mechanism rather than the intention. A hard rule that size never changes after a loss removes the entire class of decisions the state is worst at. And a physiological trigger — the urge to act arriving before the setup does — sends the trader away from the screen for twenty minutes, which costs nothing and removes the moment. There is a second-order benefit to writing the rules this way, and it is the one that makes them stick. Each rule replaces a decision with a habit, and habit is the system that survives stress. After enough repetitions the response to a third loss is not a debate about whether to halve the size, it is the halving; the trader has moved the decision to a state that can make it. That is the same move the mastery rung makes with pre-commitment, arrived at from the other direction — not by making the trader stronger in the moment, but by arranging for fewer moments to need strength. • Sleep is a position input: below the threshold, trade the smallest unit or not at all. • Three losses halve the unit until two planned trades have been followed. • A daily loss limit ends the session — closing the platform is the mechanism. • Size never changes after a loss; that removes the decision the state handles worst. • The urge to act before the setup appears sends you away for twenty minutes. • Write each rule when the state is absent, so the hot state only has to recognise it. A rule whose trigger requires judgment — “if I feel off” — is not a rule, because the state that has to recognise itself is the state that is unreliable. Triggers have to be countable: three losses, a limit, a clock, a written setup that is either on the list or is not.

Sleep is a position-sizing input

The state that precedes a bad decision is not purely psychological, and some of it is measurable. Short sleep reliably degrades risk assessment and impulse control in laboratory settings, and the effect shows up in real books: studies of traders and of gambling behaviour find that sleep-deprived participants take more risk and discount the downside. Decision quality also drifts over a session — the well-documented pattern of different returns at the start and the end of the trading day is partly a story about fatigue and partly about who is at the desk. This matters because a state you cannot feel accurately is a state you cannot self-report. People are poor judges of their own impairment: after a bad night, a trader typically feels fine while performing measurably worse, which is precisely the condition under which a rule beats an intention. The practical response is not “trade less when tired”, which requires an accurate read on tiredness you do not have; it is to make a *mechanical* adjustment — half size, no new positions after a certain hour, no trades beyond the first two of the day. The broader point is that position sizing is not only a function of the market. It is a function of the decision-maker, and the decision-maker varies. A written rule that halves risk after poor sleep or a long session converts a vague state into a number, which is the only form a plan can enforce. A rule that says “when uncertain, reduce size” is unenforceable, because uncertainty does not announce itself with a number. “Half size on fewer than six hours’ sleep” does.

Tag the state, then measure it

This lesson’s rules are written for states that are hard to observe from the inside, and the audit that makes them checkable is the one the playbook lesson already runs on trades: a tag on every entry, and an expectancy computed per group. The tag needs to be as crude as possible, because a sophisticated self-assessment taken in the moment is exactly the measurement the state corrupts. Three fields do the work. **Sleep**, recorded as a band rather than a figure — under six hours, six to seven, seven or more — because a precise number invites precision the data does not have. **Session load**, meaning whether the entry was taken in the first hour of screen time or after several, and whether it followed a loss. And **state**, a two-value field: normal, or one of the named markers this lesson identifies. A trader who records those three with every trade has, after fifty trades, three groups with whichever expectancy each carried, and the result is not a matter of opinion. What the measurement usually shows is worth anticipating, because it dictates how the rules should be written. The effect of a bad state is rarely a uniform degradation across all trades; it shows up in a specific behaviour — larger size, earlier entries, stops moved, positions held past their invalidation — so the tag interacts with the plan rather than replacing it. That is why the useful presentation is a **two-way table**: expectancy by state, and within each state, the fraction of trades that followed the written plan. A state that halves the plan-adherence rate while leaving size unchanged is a different problem from one that doubles the size within the same adherence, and the responses are correspondingly different: the first is a rule about entries, the second about sizing. Without the table, both look like the same vague observation that bad days are bad. Two cautions keep the exercise from becoming another way to feel bad about losing trades. The first is that the sample is small and the classes are confounded: poor sleep is more likely after a loss, a loss is more likely in a volatile market, and a volatile market produces more losses — so the causal story the table suggests must be treated as a hypothesis rather than a verdict, and the rule should be one that would be cheap even if the effect were absent. The second is that the purpose of the tag is not to grade the trader but to place the rule: a marker with no measured cost does not need a rule, and a marker with a large cost deserves a mechanical one. That is the discipline this whole subject is built on — measure the behaviour, then constrain it — and the state tag is the only version of it that can be applied to the person doing the trading. The two-way table, with a worked reading — Normal state: Expectancy +0.31R · 92% of trades followed the plan · Under six hours of sleep: Expectancy +0.04R · 71% followed the plan ← · After three losses in a session: Expectancy −0.18R · 55% followed the plan, and size rose ← · The rule the table justifies: Half size under six hours; the session ends after the third loss The connection to the policy lesson is direct: a marker with a measured cost converts into a line in the risk policy, and a marker without one converts into nothing. That is how this subject avoids becoming a list of things to worry about — the audit keeps only the items that cost money.

Sources

• Stress prompts habit behaviour in humansSchwabe & Wolf (2009), Journal of Neuroscience
• Physiology of real-time financial risk processingLo & Repin (2002), Journal of Cognitive Neuroscience
• A multi-lab preregistered replication of the ego-depletion effectHagger et al. (2016), Perspectives on Psychological Science
• Beliefs about willpower determine the impact of depletionJob, Dweck & Walton (2010), PNAS