Daily loss caps
Set a realized-loss ceiling per session. Once hit, the scanner stops firing new candidates and marks the account cooled off until the next session boundary.
Every trader knows their own worst habits — the revenge trade after a loss, the tired lunchtime scalp, the size-up when the day is already green. Sentio's discipline layer turns those habits into hard-coded gates. Write the rule once, in plain English. If it trips, the setup is silenced and the reason is displayed. Keeping what you've made is the point.
Every rule below is a real primitive you can configure per-model or session-wide. Chain them together to encode the way you already know you should trade — but sometimes don't.
Set a realized-loss ceiling per session. Once hit, the scanner stops firing new candidates and marks the account cooled off until the next session boundary.
Silence any model during hours you have historically traded badly — lunch chop, London-to-NY handoff, post-cash close. Muted models continue evaluating but do not push alerts.
After N losses in a row, Sentio can halve your default size, gate you out of new setups for a cooldown window, or lock your directional bias until a winning trade resets the counter.
A trade closes red — Sentio can force a mandatory cooldown, or lock you to a specific direction until an intraday reset (e.g. no counter-trend entries after a loss).
Every discipline rule can be scoped to a specific model or applied to your whole session. Aggressive setups can carry tighter guardrails than your bread-and-butter model.
Counters, mutes, bias locks — every discipline state resets cleanly at your configured session boundary (default 6pm ET). Nothing bleeds silently into the next session.
Every guardrail below is a hard rule enforced in code. It doesn't nudge you, warn you, or ask if you're sure. Once the threshold trips, the affordance to take the trade is gone.
Your realized-loss cap for the session is a number you write once. When today's aggregate hits it, the scanner stops surfacing candidates. The alert bus mutes. The decision dialog won't open. The trade you would have taken at loss #4 doesn't reach your screen.
Set the window that matches your emotional recovery — 10 minutes, 30, an hour. Between the stopout and the timer clearing, no new alerts fire from any model. The revenge trade at 14:47 doesn't get to be a trade because there's nothing to click.
Consecutive losses collapse your default position size on a ladder you configure. Two losses in a row → the sizing engine cuts your contracts by half automatically. Three in a row → quarter size. The math changes the moment the second red trade closes; you don't need to remember.
Rules that trip get logged with the specific value that caused the veto and the setup that would have surfaced. You can review the session and see what the day would have looked like without the guardrail — useful for tuning the cap up or down.
Built-in models and custom ones you author both expose the same tuning surface: a small set of numeric thresholds, a replay validator, and a diff view. Turn one knob, re-run against 3000 historical bars, compare.
Every built-in model surfaces 8–20 tunable thresholds — displacement magnitude, retest window, session gates, minimum liquidity. Each one carries an inline "how Sentio interprets it" note so you can tune without guessing what the number means.
The tuning loop is not "run overnight, hope for the best." Adjust a threshold, hit re-validate, watch the trigger count + expectancy shift within seconds. The diff view highlights exactly which historical trades disappeared or newly appeared.
Once the replay diff looks right, one button pushes the new parameter set to the live scanner. Old triggered_setups stay in your journal at their original parameters — no rewriting of history, no back-fitted stats.
Traders game rules they cannot see. Sentio makes the gate conspicuous — the setup still shows up, marked as blocked, with the exact rule and value that caused it. You cannot pretend the rule did not fire, and you have a clean record of what the day would have looked like without it.
Structural pattern completes on the timeframe. Model marks the setup as candidate.
Sentio evaluates every active rule against current session state — realized P&L, consecutive losses, active mutes, bias locks.
If any rule is tripped, the setup is gated. The gate reason is displayed next to the model card in plain English.
Gated setups are still recorded in the journal for post-session review — so you can see what you would have taken without the rule.
"Halve my size after two losses in a row, and stop scanning entirely if I'm down $500 realized today. Reset at 18:00 ET."
consecutive_losses threshold: 2 → size × 0.5daily_realized_loss threshold: -500 → haltsession_reset at: 18:00 ETDiscipline is included on every plan — free forever, no card.