Decision journals: auditing your money judgment

Here's an uncomfortable question: what's your actual track record on money decisions? Not your returns — your judgment. When you felt sure the house was a great buy, how often were you right? When you were '90% confident' the job change would work out, did nine in ten such calls succeed? Almost nobody can answer, because the instrument that would keep the record — memory — is hopelessly corrupt. It rewrites predictions after the fact ('I knew that stock was risky'), grades decisions by outcomes instead of reasoning, and quietly deletes the misses. The result is that most people run the same judgment errors for decades, feeling more experienced each year while learning almost nothing.

The fix comes from poker players, intelligence analysts, and institutional investors, who all converged on the same tool: write the decision down before the outcome exists, in enough detail that your reasoning can be audited later by the only qualified examiner — future you, holding the answer key. This is a decision journal, and paired with calibration scoring, it turns 'getting better with money' from a hope into a measurable practice.

The two corruptions the journal defeats

Hindsight bias is memory's tendency to backdate knowledge: once you know the outcome, you genuinely cannot recover what you believed before — the stock that crashed becomes 'obviously overvalued all along,' and studies show people misremember their own forecasts by wide margins in the direction of the outcome. Outcome bias (poker players call the error 'resulting') is grading the decision by how things turned out rather than by the process: skipping insurance and staying healthy gets remembered as shrewd; a sound diversified investment that hit a bad year gets remembered as a mistake. Both corruptions teach exactly the wrong lessons — hindsight bias makes you feel more prescient than you are, and resulting makes you abandon good processes after bad luck and double down on bad processes after good luck.

What to write down

The probabilities are the part people skip and the part that matters most. 'I think this will work out' is ungradeable; 'I'm 80% confident this car will need under $1,500 of repairs in three years' is a claim reality can score. Attach a number to every expectation, even a rough one. The numbers will be wrong at first — that's the point. Wrongness you can measure is wrongness you can fix.

Calibration: scoring the auditor

Calibration means your confidence matches your accuracy: of all the predictions you tag 80%, about 80% should come true. Most people are badly overconfident — classic studies find that events people call 'certain' happen roughly 80% of the time, and '90% confident' predictions land closer to 70%. Grading is simple: at each review date, mark the prediction true or false, then periodically bucket your predictions by stated confidence and compare each bucket's hit rate to its label. Twenty scored predictions are enough to see your pattern; fifty make it undeniable. Most people discover they have one systematic tilt — chronic optimism about timelines, say, or excessive pessimism about market risk — and knowing your tilt is a superpower: you can start applying a personal correction factor to your own forecasts, the way a marksman adjusts for a sight that pulls left.

Reviewing without resulting

• Grade the process and the outcome separately: a good decision with a bad result gets a note that says 'rerun it the same way' — that's the review's hardest and most valuable sentence.
• Re-read your stated reasoning before looking at what happened, to reconstruct your actual prior instead of the flattering backdated one.
• Look for repeat offenders across entries: the same rushed feeling, the same missing option ('do nothing'), the same source of bad information.
• Check the emotional-state field against results — most people find one specific state (euphoria, scarcity panic, deadline pressure) accounts for a majority of their worst calls.
• Turn each confirmed pattern into one standing rule, and write the rule where the next decision will happen.

The bottom line

You can't improve judgment you can't inspect, and memory — corrupted by hindsight and resulting — makes sure you can't inspect it. A decision journal freezes your reasoning before outcomes contaminate it; probabilities make your confidence gradeable; calibration reviews reveal the one or two systematic tilts that quietly tax every choice you make; and standing rules convert those discoveries into permanent upgrades. Ten minutes per major decision, one review session a quarter — and unlike nearly everything else in finance, the returns on knowing exactly how your own judgment fails are guaranteed to be yours alone.