Win Rate vs Expectancy: Why Most Traders Optimize the Wrong Number
A retail trader walks into a signal service and asks "what's your win rate?" The service says 70%. The trader signs up.
A different trader walks into a different service and asks the same question. Answer: 45%. The trader walks out.
This is one of the most expensive mistakes in retail trading. Win rate, in isolation, tells you essentially nothing about whether a system makes money. A 70% win rate can lose you everything. A 45% win rate can compound into wealth. The number that actually matters is expectancy — and the bizarre thing is that almost no signal service or trading course teaches it correctly.
This article walks through what expectancy actually is, why it's the right framing, and shows the exact math that proves a 45% system can beat a 70% system. If you've ever felt confused about why "high win rate" services seem to underperform their marketing, this is the explanation.
The simplest definition of expectancy
Expectancy is the average profit (or loss) you'd expect per trade if you ran the system 1,000 times. The formula is straightforward:
Expectancy = (Win rate × Average win) − (Loss rate × Average loss)That's it. Three numbers. Plug them in, get a single number that tells you whether the system makes money.
A few quick examples to anchor:
- System A: 70% win rate, average win 1%, average loss 3%. Expectancy = (0.70 × 1) − (0.30 × 3) = 0.7 − 0.9 = -0.2% per trade. Loses money.
- System B: 45% win rate, average win 4%, average loss 1.5%. Expectancy = (0.45 × 4) − (0.55 × 1.5) = 1.8 − 0.825 = +0.975% per trade. Makes money.
System B has a much lower win rate but vastly better expectancy. Over 1,000 trades, System A loses 200% of starting capital while System B gains nearly 1000%. The win rate fooled you.
Why "high win rate" services love hiding their R-multiples
The R-multiple is the ratio of average win to average loss. A 3R system means winners are 3x the size of losers. A 0.3R system means winners are one-third the size of losers.
Most high-win-rate signal services run low R-multiples. They take many small wins and accept the rare big loss. This "feels good" emotionally because most trades close in the green, but mathematically the expectancy is often negative.
Conversely, trend-following systems often have 30-45% win rates with 2-5R multiples. Most trades are small losses, but the few winners are huge. Expectancy is strongly positive, but the experience is psychologically brutal — long stretches of losses punctuated by occasional big wins.
When evaluating a signal service, never accept "win rate" without "average R-multiple" or both "average win" and "average loss". A service that won't disclose this is either ignorant of basic trading math or deliberately hiding it. Either way, walk.
A practical comparison table
Here are five hypothetical systems with the same starting capital, run 100 times each:
| System | Win rate | Avg win | Avg loss | Expectancy | After 100 trades |
|---|---|---|---|---|---|
| Scalper Pro | 80% | 0.5% | 2% | -0.0% | Breakeven |
| Day Master | 65% | 0.8% | 1.5% | -0.005% | -0.5% |
| Conservative | 55% | 1.2% | 1% | +0.21% | +21% |
| Swing System | 45% | 3% | 1.2% | +0.69% | +69% |
| Trend Follower | 35% | 6% | 1.5% | +1.13% | +113% |
Notice: the "Trend Follower" system loses 65% of the time, yet finishes with the highest absolute return. The "Scalper Pro" system wins 80% of the time and ends roughly breakeven. The system that feels the best is usually the worst.
Where transaction costs change everything
The numbers above are gross. Realistic retail trading involves 0.1-0.3% round-trip costs per trade (commissions, slippage, spread). A signal system with +0.15% gross expectancy and 0.20% transaction cost has negative net expectancy — it loses money in practice while looking profitable on paper.
This is the single biggest reason "verified" signal services with mediocre stats are still bad investments. Their gross expectancy is barely positive; after the trader's real-world transaction costs, it goes negative. The service is right that the signals have edge. The trader is right that they can't profit from them.
When computing expectancy on a system, always subtract realistic round-trip costs first. A safe assumption for retail traders is 0.2% round-trip per signal on stocks, slightly higher on crypto. If net expectancy is below +0.1% per trade after this subtraction, the system is too thin to trade profitably.
How to compute your own expectancy
If you keep a trade journal (and you should), expectancy is trivial to compute:
- Pull every closed trade from the last 30-90 days.
- Mark each as a win or loss based on whether the trade made or lost money.
- Average the percentage gain on winners. That's your average win.
- Average the percentage loss on losers (use absolute value). That's your average loss.
- Compute win rate (winners ÷ total trades).
- Plug into the formula: (Win rate × Avg win) − (Loss rate × Avg loss).
If the result is negative, you're losing money on every trade in expectation. The fix is either: take fewer trades (only the best setups), let winners run longer (increase avg win), or cut losses faster (decrease avg loss). Don't try to "increase win rate" — that's the wrong lever and usually involves smaller targets that hurt expectancy.
Why expectancy matters for evaluating signal services
If you're paying for signals, ignore the headline win rate entirely. Ask three questions instead:
- What is the average winning trade return (in % or R)?
- What is the average losing trade loss (in % or R)?
- What is the net expectancy after typical retail transaction costs?
Any service that can't or won't answer these is selling you marketing. The math underlying their signals is either unknown to them or hidden from you. Either way, you can't make an informed decision.
The SultraxAI dashboard exposes all three. Currently: average win at the 1h horizon is +0.89% on winning BTC signals, average loss is -0.72% on losing ones, net expectancy is +0.16% per signal before transaction costs. Net of 0.2% round-trip retail costs: roughly breakeven. We publish this because it's the honest answer, even though it doesn't sell as well as "54.8% win rate."
The boring truth
A profitable trading system has positive expectancy net of transaction costs. That's the only fact that matters. Win rate is a sub-component of expectancy and a misleading proxy when used alone.
The reason this is hard for most retail traders is that the systems with the best expectancy also tend to feel the worst — long losing streaks, occasional big wins. The human brain prefers the dopamine of frequent small wins, which is precisely why low-expectancy high-win-rate systems are so easy to sell.
Discipline in trading isn't "stick to your stops." It's "trust the math over the feeling." Once expectancy is positive and tracked, every trade is just one sample. The next 1,000 trades will fluctuate. The 10,000 after that will converge to the expectancy. Your job is to keep showing up while the law of large numbers does its work.
Try the platform (free)