72.5% win rate—so why am I still losing money?
My contract-trading program ran live for 933 trades, with a 72.5% win rate. Sounds high, but it still lost money.
Breaking it down makes things clear (in R, where 1R = the planned stop-loss amount for one trade):
- Winning trades earned an average of just +0.17R
- Losing trades lost an average of −0.97R
I earn a little when I win, but lose almost the full stop-loss amount when I lose. With this risk/reward ratio, I’d need a win rate of about 85% just to break even. 72.5% is nowhere near enough.
Here’s an even more sobering figure: 82% of losing trades never had an unrealized profit of even 0.25R between entry and stop-out. In other words, it’s not that they “went the wrong way” or that I “took profits too early”—they moved against me as soon as I entered.
So what if I widened the stop-loss and let winning trades run longer? I ran 36 sets of counterfactual replays: widening the trailing stop by 1.5–6 times, turning off the fast trailing layer, and testing these against different initial stop-losses. Not one set turned losses into profits. In fact, the wider the stop, the more I lost, because winning trades simply didn’t run far to begin with.
On top of that, fees accounted for 31% of the losses.
A few takeaways:
1. Win rate is the easiest metric to mislead you. A high win rate with a low risk/reward ratio is often a “picking up pennies in front of a steamroller” setup.
2. Start by looking at a strategy’s expectancy: win rate × average win − loss rate × average loss, then subtract costs.
3. Exit parameters can’t save a signal that gets the direction wrong. I spent a lot of time tuning exits, only to find that the problem was with the entries.
What risk/reward ratios have you seen in high-win-rate strategies?
#合约交易 #QuantitativeTrading
This is not investment advice.
My contract-trading program ran live for 933 trades, with a 72.5% win rate. Sounds high, but it still lost money.
Breaking it down makes things clear (in R, where 1R = the planned stop-loss amount for one trade):
- Winning trades earned an average of just +0.17R
- Losing trades lost an average of −0.97R
I earn a little when I win, but lose almost the full stop-loss amount when I lose. With this risk/reward ratio, I’d need a win rate of about 85% just to break even. 72.5% is nowhere near enough.
Here’s an even more sobering figure: 82% of losing trades never had an unrealized profit of even 0.25R between entry and stop-out. In other words, it’s not that they “went the wrong way” or that I “took profits too early”—they moved against me as soon as I entered.
So what if I widened the stop-loss and let winning trades run longer? I ran 36 sets of counterfactual replays: widening the trailing stop by 1.5–6 times, turning off the fast trailing layer, and testing these against different initial stop-losses. Not one set turned losses into profits. In fact, the wider the stop, the more I lost, because winning trades simply didn’t run far to begin with.
On top of that, fees accounted for 31% of the losses.
A few takeaways:
1. Win rate is the easiest metric to mislead you. A high win rate with a low risk/reward ratio is often a “picking up pennies in front of a steamroller” setup.
2. Start by looking at a strategy’s expectancy: win rate × average win − loss rate × average loss, then subtract costs.
3. Exit parameters can’t save a signal that gets the direction wrong. I spent a lot of time tuning exits, only to find that the problem was with the entries.
What risk/reward ratios have you seen in high-win-rate strategies?
#合约交易 #QuantitativeTrading
This is not investment advice.