What hindsight bias is
Psychologist Baruch Fischhoff described the effect in a 1975 study and called it "creeping determinism." People who were told how a historical event ended rated that ending as more likely than people who were not told. They also believed they would have seen it coming.
The bias works in two ways:
- Inevitability. The outcome now seems like it was bound to happen.
- Memory distortion. You recall your earlier forecast as closer to the outcome than it was.
Both feel like learning. Neither is.
How it shows up in investing
| Situation | What hindsight says | What was true at the time |
|---|---|---|
| A stock doubled after a spin-off | "The parent was obviously cheap" | The value depended on how the separated company would be valued, which was uncertain |
| A merger broke on antitrust grounds | "That deal was never going to close" | Many holders priced a high chance of closing, and the ruling could have gone either way |
| A biotech fell on trial data | "The phase 2 data was weak" | The same data supported a reasonable chance of success |
| A market fell sharply | "The warning signs were everywhere" | Warning signs are always present; most do not lead to a fall |
| A position you sold kept rising | "I knew it was a winner" | Your notes may show you sold because you doubted it |
In each case the outcome rewrites the story. The evidence that pointed the other way fades from memory, and the evidence that fits the result is remembered as decisive.
Why it matters
It corrupts the review
A post-mortem is supposed to separate good decisions from good outcomes. Hindsight bias merges them. A sound decision that lost money looks foolish, so you may drop a method that works. A careless decision that made money looks skilled, so you repeat it. The same trap appears in survivorship bias, where only the visible winners shape the lesson.
It inflates confidence
If the past looks predictable, the future seems predictable too. That leads to larger positions, tighter stops and less attention to the outcomes you did not expect. It works alongside confirmation bias, which steers attention to evidence that supports the view you already hold.
It distorts how you judge others
Analysts, managers and boards get blamed for not foreseeing outcomes that were genuinely uncertain. An investor who thinks a CEO "should have known" may overpay for the next management team that promises certainty.
Decisions versus outcomes
The core defense is to judge a decision on what was knowable when it was made.
Was it a bad decision? Not on its own. With 60/40 odds and those values, the expected value was above the price. The question for the review is whether the 60% estimate and the two values were reasonable given the evidence at the time, not whether the outcome was a loss. If you made 20 such decisions with honest estimates, you would expect to lose about 8 of them.
This is the logic of probability-weighted valuation, and it only works if the probabilities were written down before the event.
How to correct for hindsight bias
- Write a dated thesis. Before you act, record the thesis, the catalyst and its date, your probabilities and the evidence. How to write a dated investment thesis covers the format.
- Keep a decision journal. A decision journal captures how confident you were and why, in your words at the time.
- Name what would change your mind. If you list in advance the facts that would prove you wrong, you cannot later claim you saw them coming.
- Run a pre-mortem. Imagine the position failed and write the most likely reason. A pre-mortem puts the losing case on paper before the result.
- Review against the record. After the catalyst, run a post-mortem using only what you wrote, then compare it with what happened.
- Classify each result. Good decision, good outcome; good decision, bad outcome; bad decision, good outcome; bad decision, bad outcome. Only the decision column tells you what to change.
| Review question | Answer from memory | Answer from the record |
|---|---|---|
| What did I expect? | Usually close to what happened | What I actually wrote |
| How confident was I? | Usually more than I was | The probability I stated |
| What was the risk? | The one that occurred | The full list I made |
| Was it a good decision? | Judged by the result | Judged by the reasoning and evidence |
The record is also what makes research checkable. Dombey publishes each thesis before its catalyst and keeps it in the research archive, so the original case can be compared with what followed. Tools for keeping your own records are covered in the best stock research tools for individual investors.
Hindsight bias in the research you read
The bias does not stop at your own notes. It shapes much of what investors read after a big move.
- Retrospective articles. After a stock collapses, explanations appear quickly and sound certain. They are written by people who know the ending, and they rarely mention the evidence that pointed the other way.
- Track records without dates. A service that shows past calls without the date and price at which each was published lets the reader, and the author, fill in the timing with hindsight. A call is only checkable if it was recorded before the outcome.
- Selective memory in public commentary. Commentators who warned of many risks will be remembered for the one that happened. The warnings that did not come true fade from the record.
- Backtests. A strategy designed after looking at past data carries hindsight in its rules. The rules fit the past because they were chosen with the past in view.
The same test applies to research you pay for. Ask whether each recommendation was published with a date, a price and a catalyst, and whether the losers are kept in the record alongside the winners. A dated archive is the only defense a reader has against an author's hindsight.
Where this goes wrong
- Writing the thesis after buying. A thesis written after the price moves is already shaped by the move.
- Vague probabilities. "Likely" can be remembered as 55% or 90%. Use numbers.
- Reviewing only losses. Wins need review too, or luck gets counted as skill.
- Editing old notes. Keep the original. Add dated updates below it.
- Over-correcting. Some outcomes really were predictable, and some mistakes were real. The goal is an accurate review, not a forgiving one.
The bottom line
Hindsight bias makes known outcomes look predictable and rewrites your memory of what you expected. For investors, it turns the review step into a source of false lessons and excess confidence. The correction is simple and unglamorous: write down your thesis and probabilities before the event, keep the record unchanged, and judge decisions by what was knowable at the time. Outcomes still matter, but separately.

