Navigate the rules, daily drawdown limits, and consistency guidelines of prop trading firms. Discuss how to pass funded account challenges using scalping strategies.
An 8% target with a 5% daily and 10% maximum loss looks like a reasonable ratio. In practice, the binding constraint is not the target but the variance of the path to it.
Simple framing: if you risk 0.5% per trade with a 50% win rate and 1.5R average winner, the expectation is +0.25% per trade. Reaching 8% needs about 32 trades on average. The question is how often a losing streak in those 32 trades reaches the max-loss floor first.
Ways traders lower that risk:
Smaller risk per trade early, scaling up only after a buffer exists.
No time pressure: many firms now have no maximum duration. Use it.
Avoid "catch-up" sizing after a red day, which is the classic failure path.
Stop at target. Trading past it in phase 1 adds risk with no benefit.
Running a simple Monte Carlo of your own stats against the rules is the most honest way to know your real pass probability.
What risk per trade do you use in a challenge, and do you change it once you have a buffer?
FXS Prop Desk wrote: the expectation is +0.25% per trade.
Quick check on the maths there, because it actually strengthens your point. With 0.5% risk, a 50% win rate and a 1.5R average winner, each trade averages 0.5 x 0.75% minus 0.5 x 0.5%, which is +0.125% rather than +0.25%. So the 8% target needs about 64 trades on average, not 32. Twice as many trades means more room for a long losing run before you get there. With a coin-flip win rate, a streak of six or seven losses somewhere in 64 trades is quite normal, and at 0.5% risk that's 3 to 3.5% gone. Survivable on a 10% max limit, but uncomfortable, and it's a big part of why the target is harder than it looks.
The part I would stress is how fragile that path gets when the win rate is slightly off. Using the same 0.5% risk and 1.5R winners, dropping from 50% to 45% wins takes the edge to roughly +0.06% per trade, so the 8% target is now well over a hundred trades away.
Most of us estimate win rate from a backtest or a good month, and live fills usually shave a few points off. I run my Monte Carlo at the backtest figure minus five points and only treat the result as my pass probability if it still looks acceptable.
The daily 5% almost never binds at that risk, since it takes ten straight losers in one session. The 10% maximum is the real floor, and whether it trails or stays static changes the answer more than anything on the list above.
My own answer to the question: 0.4% until the account is up 3%, then 0.6% with the buffer as the cushion. If the account drops back under +1% I return to 0.4% until it recovers.
Do people here run the simulation with a trailing floor, or just the static one the rules page shows? The trailing version gives a noticeably lower pass rate in my runs.
Fairman, your correction holds, and costs push it further still. If spread and commission come to about 0.1R per trade, that is 0.05% off each ticket at 0.5% risk, so the edge drops from +0.125% to +0.075%.
At that rate the 8% target is roughly 107 trades away. Taking about three a session, that is 36 sessions, close to two months of trading days.
Most challenge calculators I have seen leave costs out entirely. On a scalping book with small winners, the cost line can be the biggest single number in the whole expectancy sum.
The losing streak side grows with the trade count too. Over 107 coin flip trades, a run of eight straight losers shows up in roughly one sequence out of five, and at 0.5% risk that is 4% gone before the next winner.
That is still inside a 10% maximum, but it lands near the point where catch up sizing usually starts. Knowing the number in advance is what keeps me from treating an ordinary streak as an emergency.
FXS Prop Desk, does your Monte Carlo charge costs per trade, or does it use gross R? The gap between the two answers is bigger than people expect.
Many backtests use a fixed spread, often a number like one pip applied to every trade. Real spreads vary throughout the day, and the variation is exactly where scalpers lose money.
During the Asian session, the spread might be one and a half pips. At rollover, it might be ten. Around news, it might jump to five. A fixed value of one pip understates the cost of trades taken at those times, and it overstates the profit.
The impact is largest for strategies with small targets. If the average profit per trade is two pips and the actual average cost is one and a half, the backtest looks profitable when live results would show a loss.
The remedy is to use historical spread data, either recorded from a broker or taken from a tick dataset that includes bid and ask. Many platforms support variable spread testing, though it is usually not the default.
Another useful step is to test the strategy with spreads doubled. If it collapses, the edge is fragile.
What happens to your best strategy when you run the backtest with double the normal spread?
Fairman wrote: which is +0.125% rather than +0.25%. So the 8% target needs about 64 trades on average, not 32.
You're right, 0.5 x 0.75 minus 0.5 x 0.5 is +0.125% per trade, so roughly 64 trades rather than 32. It strengthens the point, as you say. It also matters on challenges that still have a time limit: at a few trades a day, 64 trades is several weeks, and the temptation to add trades near the deadline arrives exactly when the running total is behind.
The 64-trade calculation is a useful back-of-the-envelope check, but dividing an 8% target by a 0.125% average does not forecast the number of trades to a pass. It treats returns as additive and assumes the same net expectancy on every trade. If risk is a fraction of changing equity, gains and losses compound, so an 8% balance target is not identical to summing fixed percentages of the starting balance. More importantly, a challenge is a path problem: each simulated run must reach the target before a daily or maximum-loss limit, any trailing drawdown, and perhaps a deadline. A Monte Carlo should apply those rules in sequence and include costs in the trade returns. If outcomes cluster by session or regime, independently shuffling individual trades may understate streak risk; blocks of trades or session-level resampling can be a useful comparison. I would report the estimated pass rate with uncertainty and distinguish paths that time out from those that breach a limit. The win rate and average R should also be stress-tested on data not used to tune the strategy. A preplanned size change after a buffer can be modeled, but it is a different risk policy, not a guaranteed cushion.
