How can a system that wins 70% of the time still lose?
The previous lesson's formula runs in both directions. Feed it a high win rate and a poor payoff ratio and it returns a negative number — which is exactly the shape of the most seductive way to lose money in markets, because it feels like winning right up until the account statement arrives.
Worked: the 70% system
A hypothetical record of 200 trades: 140 wins averaging $120, 60 losses averaging $400.
- Expectancy = 0.70 × $120 − 0.30 × $400 = $84 − $120 = −$36 per trade.
- Long way: (140 × $120) − (60 × $400) = $16,800 − $24,000 = −$7,200 over 200 trades. Agrees: −$36 × 200.
Seven trades in ten were winners. The record loses money.
The condition, rearranged
Expectancy is zero when the two sides balance: Win rate × Avg win = Loss rate × Avg loss. Rearranged into a payoff requirement:
Minimum payoff ratio = Loss rate ÷ Win rate
At a 70% win rate: 0.30 ÷ 0.70 = 0.43. The average win must be at least 43% of the average loss. This system's ratio is $120 ÷ $400 = 0.30. Below the line at a 70% win rate, so negative. Being right more often does eventually fix it — run the same formula the other way and a 0.30 payoff ratio breaks even at a win rate of 1 ÷ (1 + 0.30) = 76.9%, and at 80% the identical $120/$400 system turns positive at +$16 a trade. But 76.9% is a long way above 70%, and as the next section shows, the habits that lift a win rate are usually the same ones pushing the payoff ratio down.
Run the same arithmetic at other win rates and the trade-off becomes a single curve: at 50% you need a payoff above 1.0; at 40% above 1.5; at 30% above 2.33; at 80% you can survive on 0.25. Every point on that curve is breakeven. Systems don't live at a win rate — they live at a pair.
Where 70%-win systems come from
Almost always from two habits that feel like virtues:
- Taking profits quickly, which converts would-be large winners into small ones and raises the hit rate. It feels like discipline.
- Holding losers, hoping for a return to breakeven, which converts small losses into large ones. It feels like conviction.
Both push the win rate up and the payoff ratio down — in opposite directions on the same curve. The scoreboard you notice ("I'm right 7 times out of 10") improves while the one that pays you gets worse.
The asymmetry that finishes it
Fourteen wins at $120 total $1,680. One $2,000 loss erases all of them. A high win rate paired with an uncontrolled left tail is a sequence problem: the order in which the trades arrive determines whether you're still trading when the tail shows up.
The rule worth memorising
A win rate is meaningless without its payoff ratio, and a payoff ratio is meaningless without its win rate. They are one statistic in two pieces, joined by the expectancy formula. Any claim quoting only one of them — in an ad, a course, a forum post — has told you nothing at all.
Try it now
- Compute expectancy for this hypothetical record: 100 trades, 74 wins averaging $95, 26 losses averaging $310. Positive or negative?
- Compute the minimum average win that would make that record break even, using Loss rate ÷ Win rate × Average loss.
- Now change a single one of those 26 losses to −$1,500 and recompute the total. Note how far a 74%-win record moves on one tail event — and why "win rate" alone belongs nowhere near a performance claim.
- Then find the tail in real prices. In the five years below, look for a session whose open sits more than 15% away from the previous close and Measure the gap. A loss that size needs no bear market, just one ordinary Tuesday morning.