Resulting: Why a Winning Trade Is Not Proof of a Good Decision
Key takeaway
- Resulting is grading a decision by how it turned out. In a game with luck in it, that grades the luck.
- Every decision belongs in one of four boxes. Your account statement can only tell the columns apart, never the rows.
- The fix is not to think harder. It is to write the reasoning down before the result exists, and to state confidence as a number rather than a verdict.
Learning pathJudge the decision, not the outcomeStep 1 of 13
Based on Thinking in Bets — Annie Duke, 2018
The habit has a name, which is most of the cure
Resulting: deciding how good a decision was by looking at how it turned out.
Annie Duke spent eighteen years as a professional poker player before writing Thinking in Bets, and the book's central observation is one that transfers to markets without modification: in any activity where luck contributes to outcomes, the outcome is a corrupted signal about the quality of the choice that preceded it.
Poker players know this because they lose with the better hand several times a night. Investors mostly do not, because the feedback is slower and quieter — but the structure is identical, and so is the error.
Why this is expensive rather than merely philosophical
If you grade decisions by results, you will learn the wrong lesson from every box on the right of that grid — including the one you should be worried about.
The same mechanism runs in the other direction and is easier to spot: a well-reasoned position that goes against you gets abandoned, and the method with it. That is the trade after the loss, viewed from the decision side.
Chess, poker, and which one you are actually playing
Duke's framing is that most people default to chess reasoning: a loss means a mistake was made, so find it. In a game with luck, that search finds something whether or not anything was wrong — and the thing it finds becomes a rule you follow afterwards.
The further right on the line, the more repetitions you need before results say anything. A single trade says nothing. Thirty trades of one method say something. This is the same argument expectancy makes with arithmetic instead of psychology.
Stating beliefs as numbers, which is harder than it sounds
Replace "it's going up" with "I'm 65% sure it goes up" — and then be willing to be held to the 65%.
Two things happen when you attach a percentage. First, you stop needing the world to agree with you in order to have been right — a 65% call that loses is entirely consistent with a 65% call. Second, and less comfortably, you make yourself checkable: if the things you call 65% come in 40% of the time, that is now visible.
A verdict
- Right or wrong, one trade at a time
- Every loss is an identity problem
- Nothing accumulates
A probability
- Scored across many calls
- A loss can be fully expected
- Calibration improves with counting
Three practices that survive contact with a real account
- Write the reasoning before the result exists. A thesis recorded at entry is the only version that has not been edited by knowing the answer. This is the entire argument for a trade journal, and it is a stronger one than record-keeping.
- Grade the process on a separate line from the P&L. Two columns: did I follow my rules, and what happened. Four combinations. The interesting rows are the ones where the two disagree.
- Ask "what else could have happened?" After a win, list the versions of the same week where it did not work. If several of them are plausible and would have hurt, the decision was worse than the outcome.
Where this framing runs out
- It can become an excuse. "Good decision, bad luck" is available for every loss, and used reflexively it protects a bad process permanently. The check: could you have written down the reasoning in advance, in that form, with those numbers? If not, it was not a good decision.
- Outcomes do matter, in aggregate. A hundred trades of losses is not a hundred instances of bad luck. Process talk is a way to read individual results, not a way to ignore a distribution.
- Calibration takes real repetitions. Fewer than fifty recorded probability estimates and you are grading noise — the same sample-size problem that shows up everywhere else in this section.
Try this week
- Before your next entry, write one sentence of reasoning and a confidence percentage.
- Take your last winning trade and list three ways the same week could have gone against you.
- Add a "followed the rules: yes / no" column to your record, separate from the result.
- At the end of the month, group your calls by stated confidence and see how often each group came in.
Common questions
What does resulting mean in Thinking in Bets?
It is the habit of judging the quality of a decision by the quality of its outcome. In any activity where luck plays a part, it makes you draw conclusions from the luck rather than from the reasoning.
How do I judge a trade if not by whether it made money?
By whether the reasoning was written down beforehand, whether the position size matched the stated confidence, and whether the exit followed the rule you set. Those three are checkable independently of the result.
Isn't the outcome the only thing that matters in the end?
In aggregate, yes — a method that loses over hundreds of trades is a bad method regardless of the reasoning. The distinction matters trade by trade, which is the level at which people actually update their behaviour.
Why should I state my confidence as a percentage?
Because it makes you gradeable across many calls rather than right or wrong on one, and because it forces the size of the position to follow from the strength of the view instead of from how the morning is going.