Work Backward From a Drawdown Limit to Risk Per Trade

[Fairman]

Work Backward From a Drawdown Limit to Risk Per Trade

You must keep your maximum drawdown under 10 percent. Your strategy has a 40 percent win rate. Over 100 trades, the probability of a run of 8 consecutive losses is about 49 percent, as computed earlier for a 60 percent loss rate. What risk per trade lets you survive such a run comfortably?

Decide how much of the limit a bad run may use. Say a streak should consume no more than 60 percent of the limit, which is 6 percent. Eight losses at risk r cost roughly 8r, so 8r equals 6 percent, and r equals 0.75 percent.

Check that the number is robust. A streak of 10 losses, which has a lower probability, costs 7.5 percent at 0.75 percent risk, still under 10. A streak of 12 would cost 9 percent, close to the limit.

How likely is a run of 12 in 100 trades at a 60 percent loss rate? Much lower, about 8 percent, but not negligible over several hundred trades.

Check the compounding effect. Eight losses at 0.75 percent compound to a loss of 5.8 percent, slightly less than 6.

Consider the profit side. At 0.75 percent risk and 0.2R expectancy, 20 trades a month earn about 3 percent.

Use the chain: limit, share used by a bad run, expected length of the bad run, and then risk.

Try this: fill in each number for your own rules and compute the risk per trade.

[FXS HongKong Scalper]

Working backward from the limit is a clean way to set risk. One thing I would add is a margin for slippage and gaps, which can make a streak cost more than 8r. Rounding down to 0.5 or 0.6 percent leaves room for that.

[Fairman]

The chain from limit to risk is a clear method. Starting from a 10 percent limit, allowing a bad run to use 60 percent of it, and estimating the bad run at eight losses gives 0.75 percent risk per trade. The streak probability checks out: with a 60 percent loss rate, the chance of at least one run of eight losses in 100 trades is roughly 1 minus e to the minus 100 times 0.6 to the eighth times 0.4, about 49 percent. So an eight loss streak is close to a coin flip over 100 trades, which makes it a sensible baseline for planning rather than a worst case.

[Fairman]

The robustness check is useful. A twelve loss streak, with a probability of about 8 percent in 100 trades, would cost 9 percent at 0.75 percent risk, close to the limit. Over 300 trades, the chance of at least one twelve loss streak rises to roughly 23 percent. So a trader planning to trade for a year or more at this win rate should consider that longer streaks are likely at some point. Reducing risk slightly, to 0.6 percent, would keep a twelve loss streak at about 7 percent. The right balance depends on how long the trader plans to operate under the limit.

[Amanhasnoname]

When I first tried this on a funded-style account with a 10 percent limit, I did the same sums and landed on 0.75 percent, then watched a bad week take me down 4.6 percent in six days. It was not even a clean losing streak, just losers mixed with two small winners, so the plain "n losses times r" estimate understated what the equity curve actually did.

What I do now is size from the worst drawdown in my own trade log rather than from a theoretical streak. I take the deepest peak to trough in R terms, which was about 14R over a year of live trading, and make sure that number times my risk stays under roughly half the limit. That pushed me down to 0.35 percent per trade, which felt timid but kept me comfortable.

I also added a rule that cuts risk by a third once I am down 3 percent from the high, and restores it only after I regain half of that. It costs a little profit, but I have not come close to the limit since.

[Fairman]

The compounding effect is a small bonus in the trader's favour. Eight losses at 0.75 percent compound to about 5.8 percent rather than 6 percent, because each loss is calculated on a slightly smaller balance. Over longer streaks, the difference grows: twelve losses compound to about 8.6 percent rather than 9. That's the protective effect of fixed fractional sizing, as discussed in the resizing thread. It's small, but it means the simple multiplication slightly overstates the damage. For planning, using the simple figure gives a small extra margin of safety.

[Fairman]

The profit side shows the trade off clearly. At 0.75 percent risk and 0.2R expectancy, 20 trades a month earn about 3 percent. That's a solid return, but it's achieved slowly and with discipline. A trader who wants 10 percent a month would need much higher risk, which would break the drawdown limit in a normal losing streak. Filling in the chain with your own numbers, as the post suggests, shows what return is realistic within your limits. Has anyone here worked through this calculation and found their current risk was too high for their drawdown limit?