Sleep and Trading Performance: A Go/No-Go Check

A single short night reliably increases subjective sleepiness and weakens sustained attention — slower responses, more attentional lapses. Broader or more extensive sleep loss is linked to a wider set of executive-function effects, but that evidence base is dominated by total sleep-deprivation protocols, not one ordinary short night. The evidence reviewed here does not establish a universal number of sleep hours, or a universal Karolinska Sleepiness Scale (KSS) score, at which every trader should stop trading. What the evidence supports is narrower and more useful: a repeatable pre-session go/no-go check — rate current sleepiness on a fixed scale rather than a guess, name one or two decisions worth watching more closely today, and route the session into whatever normal, reduced, or no-new-entries state your existing plan already defines, the same way a risk-state check routes a session after a loss.

This is not the daily trading routine itself. The daily trading routine covers the full pre-session, live, and review sequence and treats “confirm risk state” as one generic step. This page exists because sleep loss is common enough, and its evidence specific enough, to deserve its own check inside that step — not a rewrite of the routine, and not a substitute for it.

Why sleep needs its own check, not a generic fatigue note

Most process-eligibility conversations in trading collapse into one vague category: “am I in the right headspace to trade.” That framing is too broad to act on. Sleep loss has a better-documented signature than general tiredness or a bad mood, but the evidence differs according to the type and severity of the sleep-loss protocol studied, not according to a linear reading of the exact hours lost. A single short night is most reliably linked to increased sleepiness and weaker sustained attention; broader or more extensive sleep loss, particularly total sleep deprivation, is linked to a broader set of executive-function effects. Treating every “I didn’t sleep well” as identical — or assuming worst-case research about extended sleep loss applies equally to one ordinary short night — both throw away information a trader could otherwise use.

Treating “I didn’t sleep well” the same as “I’m distracted today” also loses information. A distraction is often solvable by removing the distraction. A night of missed sleep is not solvable in the next ten minutes — the only lever available before the session is deciding how much of today’s decision-making to expose to a known, evidence-bounded change in alertness, and how.

What the evidence actually says — and what it does not

Sleep-loss research does not support a single claim like “sleep loss always makes traders take more risk,” and it does not support treating one short night the same as an all-nighter. Research on a single partially restricted night supports a narrower set of effects than research dominated by total or more extensive sleep deprivation — a difference in study protocol and evidence scope, not a clean dose-response curve running from a few hours lost to many.

One night of shortened sleep

A 2024 meta-analysis pooling 44 studies of one night of sleep restriction (roughly 2–6 hours of sleep opportunity, compared with a full night) found a large, consistent increase in subjective sleepiness and reliable impairment of sustained attention — slower responses and more attentional lapses. The same meta-analysis did not find significant pooled effects on working memory or inhibitory control.¹ That is a narrower result than “sleep loss impairs executive function”: for a single short night specifically, the strongest and most consistent evidence is sleepiness and attention, not judgment or impulse control. Within that 2–6-hour range, the degree of sustained-attention impairment was not significantly associated with exactly how many hours of sleep were lost — the finding is that restriction in this range impairs attention, not that losing more hours predicts proportionally worse attention.¹

Broader and more extensive sleep loss

A larger 2025 meta-analysis screened 79 eligible publications comprising 80 independent studies (209 effect sizes, 1,983 healthy participants) and found broader impairment across core executive functions — working memory, inhibitory control, and cognitive flexibility — across the sleep-loss protocols studied.² Its evidence base, however, is dominated by total sleep-deprivation protocols: 73 of the 80 studies used total sleep deprivation (staying awake through the night or longer) and only 7 used partial sleep restriction — not the kind of single short night most traders will actually experience. Extending its findings to an ordinary short night goes beyond what the underlying studies support.

Read together, these two meta-analyses describe a real distinction rather than one sleep-loss effect: an ordinary short night is best supported as a sleepiness-and-attention problem, while broader judgment- and control-related effects are better evidenced for more severe or extended sleep loss than for a single missed night. A 2026 chapter synthesizing this literature reaches a consistent conclusion: sleep loss does not exert a uniform effect across every cognitive domain, and attention, vigilance, and alertness are among the functions most consistently affected.³

Decision-making specifically

A 2025 review in Sleep Medicine Reviews argues that decision-making effects are not well captured by a single global “better or worse” behavioral outcome: conventional outcome measures can obscure changes in the underlying cognitive processes or strategies a person uses to decide, even when sleep loss has changed them.⁴ It is a focused peer-reviewed review, not a meta-analysis — it does not pool decision-making studies into one effect size. A separate 2025 scoping review of the sleep-deprivation and decision-making literature found a heterogeneous picture consistent with that caution: many included studies reported increased risky choices after sleep loss, but the direction and size of the effect varied with task type, gender, and whether sleep loss was voluntary or involuntary — and studies using economic decision-making tasks specifically showed weaker, less consistent impairment than other task types.⁵ A scoping review maps the range of findings in a literature rather than pooling them into one effect size the way the two meta-analyses above do; read both sources together as evidence that decision-making effects are inconsistent, context-dependent, and not reducible to a single outcome measure — not as a stronger or more precise result than the meta-analytic evidence on sleepiness and executive function.

What trading-specific evidence shows — and its limits

The clearest trading-specific evidence sits at the population level, not the level of one trader’s one night. Research using U.S. retail brokerage data found that investors located where the local sunset falls later — used as a proxy tied to circadian timing and sleep disruption, not a direct measurement of any individual’s prior-night sleep duration — earned lower abnormal returns on their trades on average; further analyses in the same research suggest impaired investor attention as a channel.⁶ The cited research is currently available as NBER Working Paper 33477 and, as of 2026, has not yet been published in a peer-reviewed journal, so it should be treated as suggestive population-level evidence rather than a settled finding. It also does not establish that any specific trader will trade worse after any specific bad night — a large-sample average effect and an individual causal claim are different kinds of evidence.

None of this evidence supports a rule like “under six hours means don’t trade,” and none of it supports assuming a single poor night carries the same weight as a night with no sleep at all. What it supports is running a deliberate, evidence-bounded check before the session rather than assuming yesterday’s readiness carries over — and being precise about which claims that check can and cannot make.

One bad night is not the same signal as chronic sleep debt

A single night of poor sleep and a multi-week pattern of short sleep call for different responses, and folding them into one category is a common mistake.

Pattern	What it typically signals	What it does not tell you
One acute poor night (illness, travel, a late event)	A short-term increase in sleepiness and weaker sustained attention — the effects best supported by single-night research	Whether judgment, impulse control, or other executive functions are impaired tonight, or whether tomorrow will look the same
Recurring short sleep across several sessions	A standing condition that single-night research doesn’t address — repeated or chronic partial restriction is a separate protocol from both the one-night restriction and total-sleep-deprivation evidence cited elsewhere on this page, with its own, uneven recovery pattern⁷	Which specific session or trade it caused
Poor sleep plus a known personal trigger (a loss the day before, a schedule change)	A compounding-risk day where more than one condition from your plan may already apply	That sleep is the dominant cause rather than the other trigger

The acute case is a short-term state for which this same-day check is designed — best documented as elevated sleepiness and reduced attention — but the rating itself does not establish how quickly any affected measure will recover. Single-night research does not tell us exactly how much one recovery night restores, or whether every affected measure recovers on the same schedule — so “acute” should not be read as “resolved by tomorrow regardless.”

The chronic case is a different problem, and this page does not offer medical guidance on it. Recovery from accumulated sleep loss is heterogeneous across outcomes: reviewed evidence on recovery from chronic sleep restriction shows that subjective sleepiness, mood, vigilance, and other performance measures can recover at different rates, and one or two recovery nights should not be assumed to restore every measure to baseline once sleep debt has accumulated over time.⁷ That is a reason to treat a recurring short-sleep pattern as a standing condition worth raising at scheduled review — a single day’s go/no-go check will keep flagging it without fixing anything. If sleep debt is recurring, log it and raise the pattern at your next scheduled review, the same way a repeated rule deviation would be, rather than trying to solve it by running a stricter version of the same daily check.

A pre-session sleep go/no-go check

Use this inside the “confirm risk state” step of your existing routine. It has three parts: a fixed rating, a personal trigger check, and a routing decision — in that order, so the outcome doesn’t get argued with after the fact.

1. Rate sleepiness on a fixed scale, not a feeling

A vague self-assessment (“I feel okay”) is easy to talk yourself out of once the market opens. The Karolinska Sleepiness Scale (KSS) is a 9-point subjective measure of current sleepiness, from 1 (“extremely alert”) to 9 (“very sleepy, great effort to keep awake, fighting sleep”), originally described against EEG and eye-movement measures of sleepiness in active, awake subjects,⁸ with subsequent studies relating KSS ratings to objective performance and EEG measures of sleepiness.⁹ It is a subjective measure whose scores have shown relationships with behavioral and physiological indicators — it is not an objective measure, and it was not developed or validated as a trading-readiness instrument, a safety certification, or a universal threshold for a “normal,” “reduced,” or “no-new-entries” trading day. Whatever score you use to separate those states is a threshold you set for your own plan, not a research-derived cutoff.

2. Name the one or two decisions worth watching today

From your own trading records, identify which specific decisions you have reason to watch more closely today — late entries, missed invalidation levels, and slower recognition of a rule violation are common candidates to check. These are not sleep-research-derived failure modes: no cited study measured these specific trading errors. They are trader-defined process metrics that can be logged consistently to test whether they co-occur with higher sleepiness in your own records. This is not a claim that sleep loss reliably affects these particular decisions for you; it is a way to direct attention somewhere concrete instead of relying on a vague sense of “being careful.” You are not diagnosing today’s session from this list — you are deciding what to watch for if you trade.

If you want to sharpen this over time, log it prospectively rather than reconstructing it from memory or from win/loss outcomes: record the day’s rating, and afterward record process-quality signals — rule adherence, missed checklist items, invalidation levels recognized late, entry-timing errors, or other predefined behavioral markers — rather than P&L or whether the session was a win or a loss. A profitable session with a late entry and a losing session with clean execution are different kinds of evidence about the same process question, and neither is settled by the outcome alone. Enough consistent logs can show you an association worth watching between your rating and a specific process metric; they do not establish that sleep caused the result, and there is no fixed number of sessions required before a pattern is worth taking seriously — use the same judgment you would for any other self-observed trading pattern.

3. Route the day using your plan’s existing states

The rating and the trigger check are measurements — a repeatable subjective-sleepiness reading and a documented personal pattern. Which of three outcomes they map to is a threshold your own plan sets in advance, not a validated trading-action threshold from research. Combine them into one of three outcomes, using states your plan should already define rather than inventing new ones for this page alone:

Normal session: your predefined eligibility rule classifies today’s rating and markers as normal — proceed under the plan as written.
Reduced session: your plan already maps this rating-and-marker combination to a reduced-exposure or narrower-setup state — activate that state, the same one you’d use for other flagged conditions.
No new entries: your plan already maps this rating-and-marker combination, or a chronic pattern compounding with another known trigger, to a no-new-entries state — manage existing positions only, and treat the day as a scheduled-review flag rather than a trading day.

This routing step is deliberately not a new rulebook. If your plan does not yet define a reduced-exposure state or a no-new-entries state, that gap belongs in trading rules and trading guardrails — build the state once, then this check simply decides which day triggers it.

A worked example

The KSS scores and thresholds below are illustrative only, invented for this example — not research-derived or validated trading cutoffs.

A discretionary trader’s plan already defines a reduced-size state, used previously after a loss streak, and a no-new-entries state for exceeded exposure limits. After two nights of short sleep from a family emergency, the trader runs the check before the session: a KSS rating of 7 (this trader’s own, self-chosen threshold for “reduced” is 6, and for “no-new-entries” is 8 — arbitrary numbers this trader set for their own plan, not research-backed figures), plus a personal marker — late entries have shown up in this trader’s records after short-sleep nights before.

The rating and marker together cross this trader’s own “reduced” threshold but not their “no-new-entries” one. The trader applies the existing reduced-size state for the session, adds one extra pre-trade checkpoint on entry timing specifically, and logs the sleep pattern as a note for the next scheduled review, since this is now a second consecutive short-sleep night. No new rule was invented for the day, and no research claim is being made about what two short nights specifically do to this trader’s cognition — an existing plan state was applied based on a documented, self-defined trigger.

Common failure modes

Failure mode	What goes wrong	Repair
Treating any tiredness as full disqualification	Sessions get skipped on evidence too weak to justify it, and the check stops being trusted	Use a fixed rating and your own threshold, not a mood judgment
Deciding after the first trade instead of before the session	The decision gets rationalized by an early result rather than made from the sleep state itself	Complete the check before market open, as part of the existing pre-session step
Building a brand-new rule system just for sleep	The plan grows a second, sleep-specific rulebook that drifts from the rest of the guardrails	Route into the reduced-exposure and no-new-entries states the plan already defines
Ignoring a repeating pattern because each night looks acceptable alone	Chronic sleep debt keeps shaping execution without ever being named as the cause	Log recurring short-sleep nights and raise the pattern at scheduled review
Assuming the threshold from research applies to you	A published rating cutoff or study finding gets treated as a personal trading rule	Predefine your own threshold in your plan and revise it cautiously using prospectively logged process-quality signals — not P&L or an isolated losing session
Setting a threshold from remembered win/loss sessions	A handful of memorable losing sessions gets treated as proof, when the sample is small and selected after the fact	Log the rating and a process metric prospectively, before the outcome is known, and read the pattern as an association, not a cause

Where Costante fits

Costante’s Session Guardrails let a trader define boundaries in advance — a time-of-day session cutoff, an after-loss risk mode that reduces size following a loss, a daily loss limit, and a per-market re-entry limit — and these settings persist per trader, available whenever a session opens. Costante can support the parts of this go/no-go workflow that correspond to features already in the product: trades can be tagged with structured fields such as setup type, execution discipline, and failure mode, and Costante evaluates each trade’s guardrail configuration against what was actually observed — daily-loss and risk-multiple breaches, re-entry-limit breaches, after-loss sizing, and session-cutoff timing — so a trader can review deviations against the plan. Each of Costante’s guardrails is triggered by its own specific, predefined condition — a loss, a daily-loss or risk-multiple threshold, a cutoff time, a re-entry count — not by an arbitrary trader-defined readiness input such as a sleep rating; the product does not currently have a generic reduced-session or no-new-entries state that a rating like this can route into on its own.

Costante does not measure sleep, does not calculate a sleepiness or KSS score, and does not include a dedicated sleep or readiness field. The sleepiness rating itself, and the mapping from that rating to a session state, remain external to Costante unless a future version of the product adds them. Costante does not automatically correlate a logged tag with later execution quality, does not decide the threshold that separates a normal, reduced, or no-new-entries day, and does not determine whether a given session is safe to trade. Those judgments, and any correlation between a logged condition and later behavior, remain the trader’s.

Frequently asked questions

Should I just skip trading entirely after a bad night’s sleep?

Not automatically. A single short night is most reliably linked to increased sleepiness and weaker sustained attention, not to a proven, universal decline in judgment or impulse control — and decision-making research specifically shows a heterogeneous picture, with economic decision-making tasks showing weaker and less consistent impairment than other task types.⁵ A fixed rating plus your own known trigger points is a better basis for the decision than either ignoring the night or skipping by default.

Does one bad night affect trading the same way as pulling an all-nighter?

No, and treating them the same overstates the evidence for an ordinary short night. The strongest, most consistent evidence for a single night of shortened sleep is increased sleepiness and weaker sustained attention.¹ Broader effects on executive functions — working memory, inhibitory control, and cognitive flexibility — are supported by the wider sleep-loss literature, including a 2025 meta-analysis whose evidence base is dominated by total sleep-deprivation protocols rather than one short night.² Repeated or chronic sleep restriction is a separate protocol again, with its own literature on uneven recovery.⁷

How many hours of sleep is too little to trade?

The cited research does not provide a validated sleep-hours or KSS trading cutoff. If you choose to use a threshold operationally, predefine it in your own plan and treat it as a personal workflow rule — not a research-validated boundary. Use a fixed scale like the Karolinska Sleepiness Scale, and revise the threshold cautiously using consistently and prospectively logged process-quality signals from your own sessions, not a round number borrowed from an unrelated context.

Is this the same as a fatigue check later in the session?

No. This check is a pre-session eligibility decision made before the first trade. A pressure event or fatigue that develops during the session is covered by the daily trading routine’s live checkpoints, which respond to what happens after the session has already started.

Sources

Costante provides educational workflow tools, not financial advice. Trading involves risk.