How the simulator works

Risk per position, daily risk, remaining maximum loss, absorbable losses, distance to target, required R, expectancy and risk budget depend on no random draw. Same inputs, same result, always.

Absorbable losses don't come from a division. The engine replays a run of losses at the model's worst outcome, costs included, on the account's real rules, until a rule fails. That's what makes the figure right for a trailing floor, which can rise during a losing position measured on equity. The loss that lands exactly on the floor causes failure and isn't counted.

Expectancy: sum of R multiples weighted by their probability, minus costs expressed in R. For an entered distribution, the simulator also gives its variance and standard deviation.

Probability of passing, failure causes, times, distributions of result and drawdown, funded survival and withdrawals are frequencies observed on simulated paths. The interface always shows the number of paths and the seed.

Each path has its own seed, derived from the simulation's seed and its rank. Paths are therefore independent, a path can be replayed alone for display, and two simulations with the same configuration, seed and model version give exactly the same result.

The analyses (sensitivity, risk curve, solver, scenarios) use the same seed for all their points: the gap between two points comes from the setting, not from chance.

Each path is played session by session, position by position. For each position: draw the outcome, R multiple, gross P&L, deduct commission and slippage, update the balance, rebuild equity during the position, update highs, drawdown, daily loss, check the rules, then the verdict: failure, pass or continue.

During a position, a winner first pulls back then goes to its target; a loser first advances then hits its stop. It's the least favourable order for rules measured on equity. Costs are realised at the close. No position is held overnight.

Static: a floor equal to initial capital minus the maximum loss. It never moves.

End-of-day trailing: the reference high only rises at the session close, at the closing balance. A gain made then lost within the same session doesn't lift the floor.

Intraday trailing: the reference high rises after each favourable move, on balance or equity depending on the setting. On equity, the advance of a position that ends up losing already lifts the floor.

A trailing floor can be locked at initial capital: once it reaches it, it stops rising.

Fixed: an amount calculated on initial capital, measured from the morning balance. Morning balance: an amount calculated on the morning balance. Morning equity: the same base, but measured on equity, including positions' troughs.

Session high: measured from the highest balance, or highest equity, reached during the session. A session that first gained is cut earlier.

The limit is checked after each position, at the position's worst moment: the session stops at the position that breaches it.

Rules are independent objects: maximum loss, daily loss, target, minimum sessions, maximum sessions, consistency. A failure overrides everything. If two limits are breached by the same position, the maximum loss is recorded.

The target only passes once the minimum number of traded sessions is reached: a target reached in session 2 with a minimum of 5 leaves the path ongoing. The trader can then continue with the same risk or stop taking any, depending on the setting.

At the close of the last allowed session, a path that hasn't passed fails for exceeding the deadline. With no deadline, a path still active at 365 sessions is counted as ongoing, not failed.

Consistency: the best session mustn't exceed a given share of total profit when the target is reached. Depending on the setting, the evaluation fails or continues.

Each funded account starts from initial capital with the same loss limits, with no target or deadline. A withdrawal is requested at a session's close when profit reaches the threshold and the time since the last withdrawal has elapsed. The trader receives the withdrawable profit multiplied by their share.

After a withdrawal, the account follows the chosen mode, and only that: balance and floor return to initial capital, balance alone returns with a trailing floor that doesn't come down, or withdrawal limited to the profit above a buffer. A withdrawal paid out stays earned; profit left on a closed account is lost.

With a program from the catalogue, each path plays the first phase, then the next if it passed, up to the funded account. Each phase restarts from a fresh account at initial capital, with its own rules and its own random stream.

The probability of reaching the funded account is the share of paths that actually passed all phases in a row. It isn't the product of each phase's probabilities.

When a position breaches several limits at once, all are recorded; the first in the order of rules is shown as the main reason.

Rules come from published, sourced versions. Unknown data is never filled with an invented value: the simulator applies an explicit assumption, shown next to the results.

A 70% probability doesn't mean you'll pass seven times out of ten. It means that, on paths that exactly follow your assumptions, seven out of ten passed. If your real win rate is lower, so is the real figure.

Ruin by maximum loss is defined here as the share of paths that hit the maximum loss before the target. Other definitions of risk of ruin exist and would give other figures.

Each main probability comes with a 95% Wilson interval. It only measures uncertainty due to the finite number of paths: it narrows as more paths are simulated.

It measures neither the model's error nor that of your assumptions. A probability of 68.4% between 67.5% and 69.3% is still wrong if the win rate entered is wrong.

To account for this, state how many positions your win rate was measured on. Each path then draws its own rate from the beta distribution of that observed rate: over 100 positions at 50%, the drawn rate falls roughly between 40 and 60% in 95% of cases. The result widens accordingly. In the full path, the rate is the same from one phase to the next: it's the same trader.

Independent positions: in reality losses cluster in difficult periods, and long streaks are more frequent than the model estimates.

Constant win rate and outcome distribution, with no change in market regime or correlation between positions.

Losses beyond the stop (gap, poorly executed stop) are only counted if you declare a share of them; by default, every loss is −1 R plus the costs entered.

By default, a loss limit is breached as soon as the account reaches the floor. Some firms only penalise going below it: the setting exists, and the difference matters when risk lands exactly on the limit.

No specific slippage around announcements, no simultaneous positions, no psychological behaviour or execution errors.

Excursion during a position follows a chosen assumption (uniform, fixed, worst case): it isn't a market measurement. Real maximum favourable and adverse excursions (MFE, MAE) aren't used yet.

Positions are played one at a time: simultaneous positions, portfolio exposure and correlations aren't simulated.

Behaviour restrictions (announcements, weekends, overnight, robots, trade copying, arbitrage) are shown with the program but never built into the Monte Carlo.

The rule sets offered are examples, unsourced and unverified. Only your own rules are binding.