Monte Carlo on accounts: odds an account hits the max drawdown
The "Odds an account hits the max DD" sub-panel replays the whole prop firm run many times with recurring purchases, block-shuffled sessions and a buying policy, and counts what happens to accounts: share dying on the drawdown, pass rate, payouts, accounts bought and funded, and the net result with its 5–95 % range.
Senzoukria · Documentation · Updated September 2026
Where to find it
- Where
- Replay → Performance panel → Prop firm simulation → "Odds an account hits the max DD"
- Defaults
- 500 runs, block length 5, seed 20260910; policy: 1 evaluation at start, 1 per 21-session period, max 50 evaluations, max 3 funded at once
- Reading
- Low below 5 % of accounts dying on the drawdown, Real up to 33 %, Structural above
- Cost
- About 0.8 ms per run on five years of sessions; 5,000 runs take about four seconds, deferred so the screen does not freeze
What it does
The two other Monte Carlo blocks answer "what chance of passing" and "where does one account's drawdown live". This one answers the question that decides a position size: over several years, buying accounts as you actually would, what share of accounts ends up through its floor? It counts accounts, not the equity curve, because the equity curve never resets and has no floor in the firm's sense.
Sessions are reshuffled in blocks so losing streaks survive the shuffle; shuffling day by day would dissolve exactly the phenomenon being measured. The dates that come out no longer mean anything, which is fine here since only accounts and their fate are read.
Deaths on the drawdown are counted separately from other causes. An account killed by a daily loss limit or a consistency rule is not attributed to the floor. The headline reads "{x} % of accounts die on the drawdown · {n} accounts simulated · {runs} runs", followed by the reading Low, Real or Structural.
Settings
| Setting | Default | What it changes |
|---|---|---|
| Runs | 500 (100–5,000, step 100) | Number of complete paths replayed; all settings are deferred so each slider stays responsive |
| Buy each period | 1 account (0–5) | Evaluations bought at every recurring deadline; 0 means never |
| Period length | 21 sessions (5–90) | Sessions between two purchases, counted in sessions because the stream has no calendar months |
| Max evaluations | 50 (1–200) | Cap on purchases over the whole period; without it a losing strategy would buy forever |
| Max funded at once | 3 (1–20) | Buying stops while this many funded accounts are held, as most firms cap simultaneous funded accounts |
Figures
- Accounts bought (median) and Funded (median): per run, the median number of evaluations purchased and of accounts that passed.
- Pass rate: share of all simulated accounts that pass their evaluation. Reached a payout: share that cash at least one withdrawal.
- Net result (median) and Net, 5 % → 95 %: withdrawals minus purchases per run. This is not the net P&L of the curve; each account cost its purchase and pays out separately.
- The policy help text is explicit: a challenge does not last five years and nobody buys a single account over that span, so the result depends on these settings as much as on the strategy.
Limits or pitfalls
An initial batch bought the same day does not diversify: accounts trading the same strategy on the same sessions pass or die together. What decorrelates is the start offset, which the recurring purchase provides; the panel exposes the recurring setting rather than a batch size.
The Low / Real / Structural labels are reading aids, not statistical tests; they exist so that a bare percentage is not interpreted either way depending on mood.
Like every Monte Carlo panel here, it shows a spread of outcomes for one strategy and one rule set. It does not rank strategies and must not be used as a selection filter, which would be overfitting with a rigorous-looking method.
Related pages
This page in other languages
Frequently asked questions
- Why does the number of accounts bought differ from the funnel above?
- The funnel replays the real session order with one account at a time. This panel applies a buying policy with recurring purchases and caps, across block-shuffled orders; the median is over those runs.
- Why is the net result lower than my backtest P&L?
- The net here is payouts received minus the cost of every account bought, after the profit split and payout rules. Many accounts pay their purchase and never withdraw, and the equity curve never charged those fees.
- What is the block length used?
- 5 sessions, the same as the Drawdown zones panel, roughly one trading week. It is not exposed as a slider in this panel.