Prop firm Monte Carlo simulation
How the simulator plays a strategy's own trades through a firm's written rules thousands of times, with every assumption it rests on and everything it cannot see.
What a measurement is
One question answered with numbers: if this strategy traded this firm's account, how often would it pass, how often would it be paid, and what would it cost?
The answer is computed from the strategy's own closed trades and the firm's rules as written on the date they were read, and every result states what it rests on.
It is a measurement of the past, not a forecast and not advice. Simulated results have limits that real trading does not: no real money was at risk, and a strategy can be fitted to the history it is measured on. Trading futures carries a risk of loss; the Risk disclosure says more.
The trades it reads
Every closed round trip, and how finely the account was watched inside it decides what can be seen.
Every closed round trip, with the instant it opened and closed and its result after commission: from a backtest in the app, from a trading platform's export, or from a CSV pasted into the free calculator. Each trade belongs to the firm's trading day it closed on, and the day turns at the firm's own clock in the firm's own time zone, daylight saving included: 17:00 Chicago time at one firm, 18:00 New York time at another.
How finely the account was watched decides what can be seen:
| Trades from | Watched |
|---|---|
| A backtest in the app | Each hour's highest and lowest equity inside every trade |
| A trading platform's export | Rebuilt minute by minute from market data |
| A CSV or typed numbers in the free calculator | At the closes only |
A backtest in the app
- Watched
- Each hour's highest and lowest equity inside every trade
A trading platform's export
- Watched
- Rebuilt minute by minute from market data
A CSV or typed numbers in the free calculator
- Watched
- At the closes only
A loss inside a trade that recovered by its close cannot be seen at closes, so a measurement watched at closes is kinder than the firm, and it says so beside the result.
The trades can be resized: three micro contracts in place of one mini, converted through the contracts' point values, with the firm's own commissions where the firm charges its own.
The rules and their evidence
Each firm's account is one ruleset with a version, and each figure in it carries the page it was read on.
Each firm's program, account size, pricing path, platform and purchase date is one ruleset with a version. Each figure in it carries the page it was read on, the words on that page and the date. A figure no page of the firm's own states is marked “not confirmed by the firm” wherever it is shown, and sources that disagree are named.
A version
Firms change their rules often, and many changes apply only to accounts bought after a date. A change is a new version with its own date; a stored measurement keeps the version it measured; an account bought before a change is judged by the rules it was bought under.
The catalog shows each firm's versions and what changed. What is modeled:
- Loss floors that are fixed, trail the end-of-day balance, trail realized profit only, or trail every new high during a trade; where each stops rising; what a payout does to it.
- Daily loss limits that close the day and those that fail the account, on closed trades or with open positions included, and schedules that change with profit.
- Consistency rules in each formula firms use: the best day against the profit, against the target (raising the target when exceeded), against the profitable days only, and the account's best day ever against the profit since the last payout.
- Minimum trading days, winning days with their dollar threshold, inactivity, and time limits.
- Contract limits in mini equivalents and scaling plans that step with profit from the next session.
- Flat-by times and windows around scheduled US releases, read from the historical calendar of each trade's own date.
- The funded account's payout rules: winning days, buffers, caps by payout number, splits, frequency, the most payouts an account may take.
Whole trading days, resampled
Attempts are built from runs of the strategy's own trading days, so a bad morning stays one morning and a losing streak stays a streak.
One history is one ordering of the trades. To see how often a strategy passes rather than whether it passed once, the measurement builds thousands of attempts from the strategy's own trading days: runs of consecutive days drawn at random from the history and joined end to end. A run that reaches the last day goes on from the first, so every day is drawn as often as any other.
A day keeps its trades in order, their times and its date, so a bad morning stays one morning, a losing streak stays a streak and a release day stays a release day. A position still open where two runs join is dropped, never carried into a day it did not trade.
L = D1/2, at least 2, at most D ÷ 4
D is the number of trading days in the history
L is stated with every result
A run's length is the square root of the number of trading days in the history, at least 2 and at most a quarter of them, and every result states it. A history of fewer than 20 trading days is resampled in runs of trades instead, laid on days of 24 hours that each stand for one trading day and its share of weekends, and the result says so. Every attempt draws from its own seeded generator: the same trades, rules and seed give the same numbers, exactly.
The rules, in the order they act
Each attempt is judged event by event, as a firm judges an account.
1A trade opens
A news window or a contract limit may refuse it, and the trader's own limits for the day (a day stop, a goal, a trade count) may skip it.
2The account is read inside the trade
Where the trade was watched, equity is balance plus the open result: a floor that trails every new high rises with it, the floor is checked, and a daily loss limit on equity is checked.
3The trade closes
The balance moves by the trade's result after commission; the floor, the daily loss limit on closed trades and an account at zero are checked.
4The day ends at the firm's cut
An end-of-day floor moves and may lock; the scaling plan steps; an evaluation may pass; a funded account may ask for a payout; inactivity and time limits are counted. A day with no trade still ends.
A daily loss limit that closes the day flattens the account at the limit and skips the rest of that day; one that fails the account ends the attempt. Where the firm closes an account at a line, it is closed at the line less one tick per contract, the slippage of a forced exit. A consistency rule that holds the pass keeps the attempt trading until profit dilutes the best day.
Past the pass: payouts and fees
An attempt does not stop at passing: it follows the funded account to the payout, and a failed account is reset or bought again.
The first evaluation is followed for up to a year whatever the money's horizon, so the share that passes does not depend on the horizon chosen. An attempt does not stop at passing. It pays activation where the firm charges it, starts the funded account and asks for a payout on the first day the rules allow, for as much as they allow, at the split in force. A failed account is reset or bought again, as chosen, over a horizon of 90 to 365 days. Where the firm index names the best account for a goal, an account with no payout in any run is left out.
The result is money
The chance of a first payout inside the horizon and how long it takes, the fees paid before it, and the net (payouts less every fee) as a spread of outcomes, counted at the firm's list prices or at the prices traders typically pay, as chosen and stated.
The interval and the stress tests
A 90% interval holds both the simulation's sampling and the doubt about the edge itself, and it held the true pass rate 91 to 92 times in 100.
The pass rate comes with a 90% interval that holds two uncertainties: the simulation's own sampling, and the strategy's edge itself, measured again on 200 resamples of the trading days, each made of runs as long as the cube root of the days, widened for how few independent stretches a short history holds. On strategies whose true pass rate is known, the interval held it 91 to 92 times in a hundred. A history of 40 days gives a wide interval, and that width is the information.
90%, over 200 resamples in runs of D1/3 days
D is the number of trading days in the history
On known pass rates it held the truth 91 to 92 times in 100
Two stress tests rerun the same attempts:
- One more tick
- One more tick of cost on every side of every trade. A pass rate that falls apart under one tick was resting on fills a live account is unlikely to get.
- A smaller edge
- An average trade 10% smaller.
How much history is enough
Under 20 trading days or 30 trades a result shows how the trades fared and gives no chances at all.
Every result says how much of the strategy its trades hold:
| History | What the result says |
|---|---|
| Under 20 trading days or 30 trades | Too little to measure chances from: the attempts would only repeat those few days, so the result shows how the trades fared as they happened and says the history is too short. |
| Under 60 trading days or 100 trades | The chances are rough and their ranges wide, and the result says so before its figures. |
Under 20 trading days or 30 trades
- What the result says
- Too little to measure chances from: the attempts would only repeat those few days, so the result shows how the trades fared as they happened and says the history is too short.
Under 60 trading days or 100 trades
- What the result says
- The chances are rough and their ranges wide, and the result says so before its figures.
Calibration: the model against its own history
The model predicts from the earlier 60% of the days and is held to what the later 40% did.
- The earlier 60%
- The model predicts a pass rate with its interval
- The later 40%
- A challenge started on each trading day, played forward
The history is split by date, 60% and 40%. Given only the earlier part, the model predicts a pass rate with its interval. A challenge is then started on each trading day of the later part and played forward on the days that actually followed. If the share that passed falls inside the predicted interval, widened by its own sampling error, the result says consistent; otherwise optimistic or pessimistic, by the side it falls on. With fewer than 8 challenges in the later part that do not overlap in time, it says there are too few to tell.
Beside every pass rate stands the history itself: a challenge started on every trading day of the source, played forward, and how many passed.
When attempts fail
Failures are counted by the rule that ended them, by the hour, and by whether the day had a scheduled release.
Failures are counted by the rule that ended them, by the hour of the firm's day they happened in, and by whether the day had a scheduled high-impact US release, beside how common such days were in the history. A strategy that fails three times as often on release days says something a pass rate alone does not.
Base rates
A measurement should be read against what firms publish about everyone who buys their accounts.
Each figure below is the firm's own, as read on Oct 9, 2026:
| Firm | Period | Passed | Paid | Read on |
|---|---|---|---|---|
| Topstep | 2025 | 16.8% of Trading Combines started | 33.3% of traders who reached a funded account | topstep.com |
| Tradeify | August 2025 to July 2026 | 17.2% of evaluations | 28.5% of funded traders | tradeify.co |
| Take Profit Trader | 2025 | 36.22% of Trading Tests | Not in the source | takeprofittrader.com |
Topstep
- Period
- 2025
- Passed
- 16.8% of Trading Combines started
- Paid
- 33.3% of traders who reached a funded account
- Read on
- topstep.com
Tradeify
- Period
- August 2025 to July 2026
- Passed
- 17.2% of evaluations
- Paid
- 28.5% of funded traders
- Read on
- tradeify.co
Take Profit Trader
- Period
- 2025
- Passed
- 36.22% of Trading Tests
- Paid
- Not in the source
- Read on
- takeprofittrader.com
A strategy measured at 60% is either much better than most of the people who buy these accounts, or measured on a history that flatters it. The interval, the stress tests and the calibration are there to tell the two apart.
What it cannot see
Live fills, a firm's review and the future stay outside any measurement.
- Live fills
- A backtest's fills are modeled (the next bar's prices, commission per contract); a live account's are not. The one-tick stress test is the measure of that gap, not a cure for it.
- Inside a trade, at closes
- A CSV of closes, or a history with no readings inside trades, cannot see a loss that recovered before the close.
- Rules enforced by review
- Hedging across accounts, trading in concert and similar rules are counted where the trades show them, never judged as a firm's reviewer would.
- A firm's decisions
- A firm may refuse a payout or close an account under terms no model can read.
- The future
- Rules change, markets change, and an edge measured on the past can fade.
The free calculator and the app
The same engine runs in the page and in the app. The difference is what it is given to read.
| The free calculator | The app | |
|---|---|---|
| Trades from | A CSV of closed trades, or trades drawn from a win rate, an average win and loss and trades per day | A strategy it writes and backtests on CME futures data back to 2010, or an imported account |
| Inside a trade | Judged at trade closes | Each hour in a backtest; minute by minute for an imported account, rebuilt from market data |
| Engine | The same engine, in the page: 5,000 attempts unless asked for more, in a few seconds | The same engine |
| Account | None needed, and nothing is uploaded | An Atlantive account |
| Also | Searches contract counts and day limits for a plan, compares every firm on one strategy, follows several accounts on one signal |
Trades from
- The free calculator
- A CSV of closed trades, or trades drawn from a win rate, an average win and loss and trades per day
- The app
- A strategy it writes and backtests on CME futures data back to 2010, or an imported account
Inside a trade
- The free calculator
- Judged at trade closes
- The app
- Each hour in a backtest; minute by minute for an imported account, rebuilt from market data
Engine
- The free calculator
- The same engine, in the page: 5,000 attempts unless asked for more, in a few seconds
- The app
- The same engine
Account
- The free calculator
- None needed, and nothing is uploaded
- The app
- An Atlantive account
Also
- The app
- Searches contract counts and day limits for a plan, compares every firm on one strategy, follows several accounts on one signal
Reproducing a result
Every attempt draws from its own seeded generator. With the same trades and the same settings the numbers are the same, to the last digit.
1The trades
Keep the trades: the same backtest, the same platform export or the same CSV.
2The settings
Read what the result states: the rules version, the sizing, the resampling unit and run length, how finely the account was watched, the prices it counted and its seed.
3The run
Measure the same trades on the same settings, and compare the numbers.
Let our AI find the best
prop firm and challenge
for your strategy.
It runs your trades through the rules of every account, then tells you which one to buy and what size to trade.
No card needed. Free credits every month.
- Most money: Blue Guardian Futures Direct 50K, $5,084
- Best payout odds: Funded Futures Family Straight to Funded 50K, 98%
- Fastest pass: Phidias Express to Live 50K, 11 days
$5,084
Median net across 2,000 simulated runs
- Evaluation
- None
- First payout
- 97%
- Fees paid
- $1,521
Median net across 2,000 simulated runs