A strategy does not have one universal probability of passing a prop challenge. It has a probability of passing a particular set of rules at a particular position size over a specified horizon. Change the daily loss limit, the way drawdown trails, or the number of trades allowed in a day and the answer changes, even when every trading signal stays the same.
This matters when you compare programs. A smaller profit target can look easier while a tighter daily loss limit makes the program less compatible with your strategy. The useful comparison keeps the strategy constant and changes the rulebook. This article develops that comparison with hypothetical accounts. The numbers explain the mechanics and are not observed customer pass rates or a ranking of current firms.
Translate the rulebook into cash boundaries
Start with a hypothetical balance of $100,000. Program A requires an $8,000 profit and permits a $10,000 total loss. Its daily loss allowance is $5,000. Program B requires only $6,000 profit, but permits a $6,000 total loss and a $3,000 daily loss. Assume both total limits are static, daily limits reset from the opening daily balance, and all positions close before that reset.
| Model rule | Program A | Program B |
|---|---|---|
| Initial balance | $100,000 | $100,000 |
| Profit target | $8,000 | $6,000 |
| Total loss allowance | $10,000 | $6,000 |
| Daily loss allowance | $5,000 | $3,000 |
| Fixed risk per trade | $800 | $800 |
| Five full losses | $4,000 | $4,000 |
Five losses cost 4% of the nominal account in either program. That is below A's daily allowance but above B's. Four full losses already cost $3,200, so B fails on the fourth loss under these assumptions. Its lower target does not remove this exposure. A strategy that trades frequently can be constrained by the daily boundary long before the total loss allowance becomes important.
Now reduce the schedule to two trades per day without changing risk per trade. Two full losses cost $1,600. Neither daily boundary can be reached from those two completed losses alone. This removes one failure mechanism in the simplified model. It does not prove that the real strategy is safe because open losses, overlapping positions, commissions and gaps may create a larger intraday decline.
Define what a loss limit actually measures
A label such as maximum drawdown is incomplete. A static floor remains at a fixed account value. A trailing floor moves upward when a specified reference reaches a new high. The reference may be end-of-day balance or intraday equity. Those are different models. An intraday profit that later disappears can raise one floor without raising the other.
Suppose the allowance is $5,000 and the account rises from $100,000 to $104,000 before closing at $101,000. A fully intraday trailing floor can reach $99,000. An end-of-day rule based on that close might set the next day's floor at $96,000. A static floor remains $95,000. The same $101,000 closing balance therefore leaves $2,000, $5,000 or $6,000 of room, depending on the rule.
Write down whether a breach occurs when equity touches the floor or only when it falls below it. Also identify whether a target requires closed positions. A simulator that awards success on an open profit but checks losses only at trade close gives the strategy an advantage that the contract may not permit. Boundary conventions belong in the test configuration, not in an undocumented implementation detail.
Use one trading process for every comparison
Keep the code version, inputs, instrument, transaction costs, position-sizing rule and trade-generating process fixed. If you lower risk only for the strictest program, you are answering a different question. That can be useful, but present it as a second experiment: the best tested sizing policy for each program. Do not mix it with the first experiment that isolates rule differences.
For a clear baseline, imagine independent outcomes of +1R and −1R, with a 52% win probability. Here R is the planned cash risk of one trade, $800. The mean result is 0.52 × $800 − 0.48 × $800 = $32 per trade before any costs omitted from the model. A $20 round-trip cost would reduce that mean to $12. A small edge can be materially altered by seemingly small execution assumptions.
When comparing models through simulation, use the same sampled trade paths for all rule sets. This pairing makes differences easier to attribute to the rules instead of random variation between two separate simulation draws. Once a program fails or passes, stop its path. Other programs can continue along the same underlying sequence until their own stopping condition is met.
Track unfinished paths separately
Even a program without an official deadline needs an evaluation horizon for analysis. You may want the chance of completion within 30 trading days, 100 trades or six months. These horizons answer different questions. A slow strategy can have a reasonable eventual success probability but a small probability of finishing within your chosen planning period.
Every simulated path should be classified as passed, failed or still running at the horizon. The three proportions must add to 100%. Do not remove unfinished paths from the denominator and then call the remaining success rate the chance of passing by the deadline. That produces a conditional probability among resolved paths, usually a more flattering number.
Consider 10,000 hypothetical paths: 4,800 pass, 2,200 fail and 3,000 remain open after 60 days. The chance of passing by day 60 is 48%. Among resolved paths, the passing share is 4,800 / 7,000 = 68.57%. Both calculations are valid, but they answer different questions. A trader planning fees and time generally needs the first number alongside the unfinished share.
Daily grouping changes the problem
A daily limit requires actual session boundaries. Five trades around midnight are not necessarily five trades in the same program day. Use the program's reset timezone, including daylight-saving transitions. Your chart timezone and the exchange session date can differ from the account's loss-limit day. Store timestamps with timezone information so the grouping is reproducible.
Resampling individual trades can break daily dependence. A losing morning and a losing afternoon may reflect the same trendless session. Separating those trades across random days can make a daily-limit strategy look safer. Sampling complete sessions or suitable blocks preserves more of that structure. It still assumes that the historical sessions are informative about the future, which needs separate stress testing.
Overlapping trades require portfolio equity rather than simply adding completed results. Two positions can each close with a small loss but have experienced a much larger combined open loss. If only closed-trade data are available, disclose that the test cannot fully validate equity-based daily rules. Increasing the number of simulations cannot recover information that the underlying data never recorded.
Two phases need conditional probabilities
For a two-stage program, the exact relationship is P(pass both) = P(pass phase one) × P(pass phase two given phase one was passed). Suppose those probabilities are 60% and 70%. The joint probability is 42%. It is not the average, 65%, and it is not automatically equal to the phase-one probability.
A fresh account resets the balance, but it does not prove independence. A changing market regime can persist across phases. The trader may change risk after passing, or the first phase may select traders who benefited from temporary favorable conditions. Using an unconditional phase-two simulation is an extra modeling assumption that should be explicit.
For a controlled stationary model with independent future draws, the multiplication of separately estimated phase probabilities is reasonable. For historical analysis, consider simulating the phases sequentially through sessions. That preserves calendar progression and can reveal a first phase that tends to finish just before a difficult market regime. Neither method guarantees how a real attempt will unfold.
Explain why paths fail
A single pass percentage hides the decision. Record the first failure event, the account value at that event and the time taken. If daily breaches dominate, inspect loss clustering and simultaneous exposure. If total loss dominates, inspect the target-to-buffer relationship and position size. If unfinished paths dominate, inspect trade frequency and the chosen horizon before simply increasing risk.
Failure categories should be mutually exclusive when reporting a first-event breakdown. If equity crosses both daily and total boundaries at the same checkpoint, apply a documented priority or report a combined category. Separately, you may analyze whether a path would ever touch each boundary. Those overlapping diagnostics should not be displayed as slices of a pie that pretends to sum to 100%.
A comparison you can reproduce
- Save the exact strategy code and input set used to create the source trades.
- Record the rulebook version, cash thresholds, reset timezone and breach conventions.
- Test the same paths and same sizing policy against each rule set.
- Report passing, failing and unfinished proportions for a common horizon.
- Repeat with higher costs, weaker trade outcomes and session-based resampling.
- Then test alternative sizing policies as a clearly separate comparison.
For each result, retain the source trade count and dates. A 52% win-rate estimate from 50 trades is much less stable than the same estimate from a broad, independent sample. The precision of the simulator is not the precision of the trading edge. A million paths generated from an optimistic sample can give a precise answer to the wrong problem.
Finally, separate passing from economic value. A higher pass probability may come with higher fees, restrictive payouts or a smaller amount that can actually be withdrawn. Rule compatibility is one part of the assessment. The expected cash value of the entire attempt, including what happens after passing, is a separate calculation.
Read the result as a compatibility test
The useful conclusion is specific: at this risk, with these trades and these boundaries, one rule set leaves more room for this strategy's normal fluctuations. It is not a claim that one company is universally easier or that a positive backtest makes a challenge safe. A lower target, a wider buffer and fewer daily trades interact. Measuring their combined effect is more informative than choosing a program from a headline percentage.