Why martingale fails a prop challenge (and bold play beats it)
Educational only, not financial advice. Simulations and backtests are models, not promises. Check every rule with your own firm.
Martingale wins most days and then loses everything. In a prop challenge with no real edge, it passes less often than the opposite approach: a few large trades sized to hit the target. Here is why, with numbers.
The setup
Take a typical 1-step challenge: a 10% profit target, a 7% static max drawdown and a 5% daily loss limit, the same headline numbers as Lark Funding’s 1-Step evaluation. Give the trader no edge: 50% win rate, 1:1 reward to risk. Every trade pays 0.05R in spread and commission (a 0.5-point spread on a 10-point stop). Allow 60 trading days and up to 3 trades a day.
Now compare four staking plans on exactly the same trades:
- Fixed 0.5% and fixed 1% risk per trade: the "sensible" plans.
- Martingale: start at 0.5%, double after each loss, reset after a win.
- Bold play: size each trade so that one win reaches the target, capped so that one loss can’t break the daily limit or the max drawdown.
Every plan stays inside the rules: no trade may risk more than what’s left before a limit.
The results (20,000 simulated challenges each)
| Plan | Pass, costs 0.05R | Pass, costs 0.10R | Avg trades (0.05R) |
|---|---|---|---|
| Fixed 0.5% | 3.9% | 0.9% | 133 |
| Fixed 1% | 20.0% | 9.5% | 63 |
| Martingale from 0.5% | 28.2% | 15.8% | 43 |
| Bold play | 36.6% | 32.8% | 3.6 |
Simulated with this site’s martingale vs bold play simulator. The zero-cost ceiling for these rules is 7 ÷ 17 = 41.2%.
Two things stand out. First, the "safe" small-risk plan is the worst, by far. Second, martingale beats fixed risk only because it bets bigger on average, and it still loses clearly to bold play. Doubling the costs barely touches bold play, but it cuts martingale nearly in half.
Why: you can’t beat a fair game, and costs make it unfair
If prices have no drift, every trade is a fair bet. A deep result in probability, Doob’s optional stopping theorem, says that no rule for when to bet or how much can change the expected value of a fair game. In a challenge that pins down your pass rate: with target T and max loss D, you pass about D ÷ (T + D) of the time whatever you do. See the real odds of passing with no edge for the derivation.
Costs break the tie. Each trade pays a fee proportional to its size, so the total fee you pay grows with the total amount you risk across all your trades. To reach +10%, a 0.5% trader has to push dozens of trades through the spread. A bold player needs two or three. With an edge that’s negative after costs, every extra trade is another small, certain loss.
This is the classic result of Dubins and Savage’s How to Gamble If You Must (1965): in a game that’s unfavourable to you, bold play, betting as much as you need or as much as you can, maximizes the chance of reaching a goal before going broke. Timid play maximizes the number of bets, and so the house’s take.
Why martingale in particular fails
- It needs many trades. Each cycle only wins back the base stake. Reaching +10% from 0.5% base bets needs about 20 winning cycles, and every trade pays costs.
- The limits cut the doubling short. 0.5%, 1%, 2%, 4%: four losses in a row already need 7.5% of room, more than the whole 7% drawdown. Streaks of four losses at 50/50 come along every few dozen trades. When the cap kicks in, the next win no longer recovers the streak.
- It hides risk instead of removing it. Martingale turns many small wins into one rare, large loss. A high win rate feels like an edge, but the average is unchanged before costs and worse after them.
The caveats (read these before you go "full YOLO")
- Daily loss limits cap size. With a 5% daily limit, bold play can’t risk the full 10% target on one trade. It risks just under 5%, so it needs a few trades. "Bold" in a challenge means as few and as large trades as the rules allow.
- Some firms ban it. Lark Funding’s help center refers to an "All-or-Nothing" prohibited strategy, and its free restart only applies if no single trade lost more than 1%. Other firms cap risk per trade or have consistency rules. A pass that breaks the rules can be voided. With a 1% risk cap, bold play falls to 20.3% in the same simulation, no better than fixed 1%.
- This is about passing, not trading well. Bold play raises your chance of passing the evaluation if you have no edge. It doesn’t make you profitable. On a funded account, the same approach would blow the account just as quickly.
- With a real edge, the answer flips. If your expectancy after costs is positive, small trades let the edge add up. In the pass-odds calculator (same rules, no time limit), a trader making +0.05R per trade passes 74% of the time at 0.5% risk and only 52% at 2%. See position sizing for prop challenges.
- The fee is the real bet. A challenge is an option you buy from the firm. A recent working paper prices a typical $500 challenge at about $221 of expected value to a zero-edge trader (Lim 2026, not peer-reviewed). Better sizing improves your odds; it doesn’t make the fee a good deal on its own.
Try it yourself
Open the simulator and change the win rate, costs or per-trade cap. You’ll see bold play’s lead shrink as the edge improves and vanish once the edge is positive.
Sources
- Dubins, L. E. & Savage, L. J. (1965). How to Gamble If You Must: Inequalities for Stochastic Processes. McGraw-Hill. [Book]
- Williams, D. (1991). Probability with Martingales. Cambridge University Press (optional stopping theorem). [Book]
- Browne, S. (1999). Reaching goals by a deadline: Digital options and continuous-time active portfolio management. Advances in Applied Probability 31(2), 551–577. doi:10.1239/aap/1029955147 [Peer-reviewed]
- Kaminski, K. M. & Lo, A. W. (2014). When do stop-loss rules stop losses? Journal of Financial Markets 18, 234–254. doi:10.1016/j.finmar.2013.07.001 [Peer-reviewed]
- Kelly, J. L. (1956). A new interpretation of information rate. Bell System Technical Journal 35(4), 917–926. doi:10.1002/j.1538-7305.1956.tb03809.x [Peer-reviewed]
- Lim (2026). The price of a funded account: An actuarial analysis of proprietary trading firm challenges. SSRN. doi:10.2139/ssrn.7178078 [Working paper, not peer-reviewed]
"Peer-reviewed" means published in a refereed journal. Working papers and books aren’t peer-reviewed. Figures from "our own tests" are our backtests and simulations, not peer-reviewed research.