Keeping A Decision Journal
Writing down the thesis, size, and stop before a trade — not after — is the only reliable way to find out later whether your reasoning was actually good.
Prerequisites: Judging The Process, Not The Outcome
A decision journal is a short, written record made at the moment a trade is entered — thesis, expected holding period, size, stop, and what would prove the thesis wrong — kept before the outcome is known, not written or rewritten afterward. The value of the journal comes entirely from that timing. A note written after a trade has already worked or failed is written by someone who already knows the answer, and memory quietly edits the reasoning to match the result, even when the person writing it has no intention of being dishonest.
Why memory alone isn't good enough
Ask a trader two months after a losing trade why they entered it, and they'll usually give you a clean, coherent story — one that sounds a lot more confident and a lot less uncertain than what they actually felt at the time. This isn't lying; it's how memory works. Hindsight smooths out the doubt that was genuinely present in the moment and replaces it with a narrative that makes sense of the outcome you now know happened. A decision journal is the only defense against this, because it captures the reasoning before the outcome exists to distort it.
What actually goes in an entry
| Field | Example entry |
|---|---|
| Thesis | "Regional bank underpriced relative to peers after a one-off legal charge; charge is non-recurring" |
| Size and why | "$500k, half of max — conviction is moderate, not high, because the legal charge's full scope isn't public yet" |
| What proves it wrong | "Next quarter's filing shows a second, related charge, suggesting the issue is ongoing rather than one-off" |
| Stop | "$34.20, below the post-charge low" |
| Expected horizon | "6–8 weeks, through next earnings" |
A scenario
A trader writes an entry exactly like the one above before buying the bank stock. Six weeks later, the position is stopped out at $34.10 after a second legal disclosure surfaces — a related matter, not a fresh one, but enough to make the market treat the original charge as part of a pattern rather than a one-off. Reviewing the journal afterward, the trader can see precisely that this was the scenario written down in advance as "what proves it wrong." The loss wasn't a surprise or a process failure — it was the pre-specified bad outcome happening, and the stop did exactly what it was supposed to do. Without the journal, the trader might remember the trade less precisely six months later — perhaps concluding, incorrectly, that they "should have known" from some signal that in reality wasn't visible at entry, and drawing an overly harsh lesson from a trade that was actually well-handled.
A decision journal's entire value comes from being written before the outcome is known — a note written afterward is not a record of your reasoning, it's a story your memory has already reshaped to fit the result.
What it's for later
The real payoff isn't any single entry, it's reading fifty of them together after a few months. Patterns show up that are invisible trade by trade: maybe every losing trade sized at "moderate conviction" or higher actually had a vaguely written "what proves it wrong" field, while the well-defined ones held up fine regardless of outcome — a sign that vagueness in the pre-trade thinking, not bad luck, is the real problem worth fixing. That kind of pattern is only visible because the entries were honest records of uncertainty at the time, not retrospective justifications.
If an entry is hard to write because the thesis feels fuzzy or the "what proves it wrong" field is vague, that difficulty is information — a trade whose invalidation condition can't be stated clearly before entry is usually a trade that hasn't been thought through enough to be sized with confidence.
Related concepts
Practice in interviews
Further reading
- Green, Managing a Trading Desk