Prop firm simulator

Before an agent takes a prop firm challenge, Trigr replays the challenge's rules on its backtest: would it have passed, how often and how fast, and which rule breaks it. The same engine judges a single strategy, a multi-strategy agent and the Propr deploy step, so every surface gives the same verdict.

Supported challenges

Trigr agents deploy to Propr, so its challenges are the presets. Rules were read from propr.xyz/rules on 2026-10-20. Any number can be changed as custom rules.

How each rule is measured

• Daily loss is measured from the realized balance at 00:00 UTC, as Propr snapshots it. Open profit at midnight is not in that balance, so giving it back the next day is not a daily loss until equity falls under the snapshot.
• Static drawdown fails at a floor set once: the starting balance minus the limit.
• Trailing drawdown fails at the equity peak minus the limit as a percentage of the starting balance (a fixed dollar step, not a percentage of the peak). The floor stops rising once it reaches the starting balance.
• Both limits are floors on equity (open positions at the mark) and fail on a momentary touch. The target is reached on equity. A bar that both passes and breaches counts as a breach.

What the simulator answers

Pass probability (Monte Carlo)

The backtest's own UTC days are resampled in blocks of 5 into 1,000 synthetic attempts of up to 365 days, each walked under the rules. It reports passed, daily-loss breach, drawdown breach and unfinished shares, the days to pass, and the pass rate inside each quarter of the history: a wide range means the result depends on the market regime you start in.

Pass probability with no edge

A strategy with no edge still passes sometimes. To measure how often, the simulator runs a second time on the same days with their average daily gain removed: every day keeps its swings and its shape, and the attempts draw the same days, under the same rules and horizon. The edge is the pass probability minus this no-edge pass probability, in points, over the attempts both runs share, unfinished ones included; near zero shows no edge for that challenge. With a daily-loss guard what-if the edge is not measured: compare the pass probability with and without the guard.

The table's drawdown / (target + drawdown) is a reference: the fair-game pass rate of a continuous path with no time limit and a static floor, 37.5% on Classic 1-Step and 25% on Turbo. A time limit, a trailing floor and the daily limit usually lower it, and losses that gap through the floor can raise it, so no edge is measured against it.

Every daily start

Each UTC day of the backtest also starts its own challenge at that day's balance. These starts overlap, so one bad month fails about thirty of them, and they are in-sample: use them to see when and why attempts failed, not as the pass rate.

Size and daily-loss guard what-ifs

sizeScale multiplies every return of the curve, and dailyGuardPctapplies the agent's own daily-loss guard (flatten at a bar close and sit out the UTC day). Both are approximations for fast feedback: leverage, liquidations and fills do not scale the way a curve does, so confirm a chosen size with a real backtest. The edge is not measured with a daily-loss guard what-if: compare the pass probability with and without the guard.

Fidelity and caveats

Every answer says what its input could show:

• day-lows: realized day-start balances and each day's lowest bar close from the full simulation; a wick inside a bar is still not seen, and Propr fails an account on a touch.
• day-start: realized day-start balances; a dip inside a bar is invisible.
• bar-close: neither; each day is measured from the previous close, open profit included.

Where to find it

• Studio: after a backtest, open the Prop firm view next to Strategy and Trades. It replays the challenge on that strategy alone: pick a challenge or custom rules, read the pass probability beside its no-edge run (not available while a daily-loss guard is set), replay the worst, median or best start against its floors, click any day of the start calendar (arrow keys move through it), and try a size or a daily-loss guard. On a phone it sits behind the Prop firm switch in the Test tab.
• Agent builder: the Prop firm challenge panel under the preview runs the same simulator on the combined curve of every strategy in the agent. That is the verdict that counts for a Propr deploy: the account takes the challenge as one book. It opens by itself when the venue is Propr.
• Propr deploy: the “Would it pass this challenge?” step checks the chosen challenge with the same engine before the agent goes live, and can shrink sizing to fit.

From an AI assistant (MCP)

trigr_list_prop_challenges lists the challenges. trigr_simulate_prop_challenge replays one on a stored backtest of an owned strategy, and propChallenge on trigr_run_agent_backtestreplays it on an agent mix's combined curve. The two tools and the option are free; only the agent backtest it rides on is metered. See prop challenges over MCP.