forex-scalping.com

Risk of ruin with tiny edges and big size

A scalping system with a small edge per trade can look safe because of the high win rate. But combined with large size relative to the account, the risk of ruin can be significant. A win rate of 65 percent with an average win of 3 pips and an average loss of 6 pips has a thin expectancy, and a single bad streak can be devastating.

The math is accessible. Given win rate, payoff ratio, and risk per trade, a simulation estimates the probability of a drawdown of a given size or of ruin over a given number of trades.

A common finding is that risk per trade must be much smaller than intuition suggests. For a thin edge, 0.25 to 0.5 percent per trade may be appropriate, not 2 percent.

Also include execution costs. A system with a small edge before costs may have no edge after them, making ruin certain given enough time.

Run the simulation with pessimistic inputs: lower win rate, higher costs. If ruin probability is meaningful, the size is wrong.

What does the ruin simulation say about the current risk per trade?

A friend of mine blew an account with exactly this profile, a 70 percent win rate with small winners and a 2 percent risk per trade, so I ran the numbers on my own system afterwards. My results were a 62 percent win rate, average win 4 pips and average loss 5.5 pips, with costs of 0.8 pip per trade. After costs the edge was almost nothing, and the simulation showed about a one in five chance of a 30 percent drawdown inside 500 trades at 1.5 percent risk.

Cutting the risk to 0.4 percent per trade brought that chance down to under 2 percent, but also made the profit very slow, which forced me to face whether the strategy was worth running at all. I ended up keeping it only on the two pairs where my spread was smallest.

I would add that the inputs should come from live fills, not a backtest. When I replaced my tested win rate with the live one, which was four points lower, the ruin estimate roughly doubled.