Thinking in probabilities
Judge decisions by process, not by one outcome. · 9 min
Predict first
A trader writes a plan before entry: stop two ATR below, target 2R, size from the stop. The trade is stopped out for −1R. Was it a bad trade?
Here are two simulated charts. On both, the same rule: buy at the close, stop two ATR below, target 2R. Decide whether the plan is sound before anything after the dashed line exists.
One trade is one draw
A method with an edge does not win every trade. It wins often enough, by enough, over a series. Each single trade is one draw from that method's distribution of results: you know the odds before entry, but not which way this draw will fall. Mark Douglas built a whole book on this idea: a trader who expects a particular trade to win will treat every loss as a personal error, while a trader who thinks in series treats it as the cost of the edge.
The arithmetic is short. For n trades with win rate p, winners of W R and losers of 1R:
Worked example
Win 40% at 2R. Expectancy is 0.4 × 2 − 0.6 = +0.2R. Over 20 trades you expect +4R, but the spread is √(20 × 0.4 × 0.6) × 3 ≈ 6.6R: wider than the edge itself. The exact binomial count says a series of 20 ends behind 25.0% of the time. Over 100 trades that falls to 9.1%. The edge did not change; the sample grew.
Edge per trade: 40% × 2R − 60% × 1R = +0.20R. Same edge every deal; only luck changes.
This deal: 8 wins, 12 losses = +4R (expected +4R).
- Chance this edge is behind after 20 trades
- 25.0%
- Middle 90% of outcomes
- −8R to +16R
Bars are exact binomial probabilities; red totals are below zero. The highlighted bar is this deal.
Process and outcome are two different questions
Grade every trade twice: was the process right (plan written, stop placed by the rule, size from the stop), and what was the result? Four boxes come out. Two are easy. The other two teach the wrong lesson unless you name them.
Put each trade in its box: judge the process, not the result.
Stop at two ATR below the swing low, target 2R, size from the stop. Stopped out for −1R.
No stop, held through a drop, price came back and the trade closed +1.5R.
Plan written before entry; target reached for +2R.
Doubled the size after three losses "to get it back". Stopped out for −2R.
The math that makes a streak feel normal
Losses come in runs even when nothing is wrong. At a 40% win rate, any five given trades are all losses with probability 0.65 ≈ 7.8%, and somewhere in 100 trades a run of five or more turns up 98% of the time. Know your number before the streak arrives; the simulator's cool-down shows it to you at your own win rate.
5 losses in a row at a 40% win rate: 7.8% for any given run of 5 trades, and 98% chance of at least one such streak somewhere in 100 trades.
After ten simulator trades, this card also checks your own longest losing run against your own win rate.
Common mistake: changing the method after a few losses, or keeping a bad habit after a few wins. Both judge by the last outcome. Change a rule only when the process review (lesson E3) or a test over many trades says so.
Check your understanding
A method wins 40% at 2R. You lose five in a row. What does that tell you about the method?
What makes a losing streak feel normal instead of alarming?
Expectancy is +0.2R per trade. What do you expect after 20 trades, and how sure can you be?
Which trade belongs in the “bad process, good result” box?
Closing round
Play five simulator rounds on the uptrend market. Before each entry, write the plan; after each exit, put the trade in one of the four boxes. Count the boxes, not the R.
Further reading: Douglas, Trading in the Zone. The explanations here are our own. Charts are simulated, not real market data, and nothing here predicts real prices.