Stock Post-Mortems: Separating Process Error from Outcome Error

Losing money on a stock is information. The question is what kind. If your thesis was sound and the world threw you a curveball, that's an outcome error — bad luck, not bad logic. If your thesis was broken from day one and the loss was the market eventually noticing, that's a process error. Treating those two the same is how prosumer investors slowly grind their edge into pulp.

This post is a framework for running the post-mortem properly. The goal isn't to make yourself feel better or worse. It's to figure out which mistakes are worth changing your behavior over and which ones aren't.

Why Outcome-Based Reviews Mislead Investors

Poker players figured this out decades ago: a hand can be played correctly and still lose. Annie Duke calls the opposite "resulting" — judging a decision by how it turned out rather than how it was made. Investors do this constantly. You sold NVDA in 2022, missed the run, and now you "should have known." Should you have? The information set in late 2022 — rate hikes, crypto-mining demand collapse, a hyperscaler capex pause — supported caution. The decision was defensible. The outcome was wrong.

The inverse is more dangerous: a sloppy thesis that prints money. You bought a meme stock on vibes, made 40%, and now you've quietly updated your priors to "vibes work." The market rewarded a bad process, which means next time you'll size up. That's how blow-ups happen.

Good post-mortems separate the decision quality from the result. You evaluate the call you made with the information you had at the time you had it.

A Framework for Auditing a Losing Trade

Walk every loser through these five questions, ideally in writing. The discipline is the point.

1. What was the original thesis, in one sentence? If you can't reconstruct it cleanly, that's already a finding — you bought without a thesis. Go back to your notes, the broker message, the screenshot, whatever you have. No reconstruction-from-memory; memory rewrites itself.

2. What had to be true for the thesis to work? This is the load-bearing question. A thesis like "Disney will recover as parks normalize" has maybe three pillars: park attendance returns, streaming losses narrow, content slate doesn't deteriorate. Write the pillars down.

3. Which pillar broke? Walk through each one with the benefit of hindsight. Did park attendance actually recover? (Yes.) Did streaming losses narrow? (Eventually.) Did the content slate hold? (Arguably not.) Now you know what killed the trade — not the price chart, the actual fundamental driver.

4. Was the broken pillar foreseeable from your pre-trade information set? This is the process-vs-outcome question. If signs of content slate weakness were visible in box office data and analyst notes you read before buying, that's process. If the slate weakened because of a strike nobody priced in, that's outcome.

5. Did sizing match conviction? A 1% position that goes to zero teaches you almost nothing. A 15% position that drops 30% teaches you whether your sizing process is calibrated to your actual edge. Most retail blow-ups are sizing errors disguised as analysis errors.

Process Errors vs. Outcome Errors: Telling Them Apart

Here's the working rule: if you ran the same decision back 100 times with the same pre-trade information, would it still have been the right call most of the time?

Process error — the decision loses money in most of those 100 runs. You missed something knowable. Examples: ignoring deteriorating gross margins because the revenue line looked great; buying a turnaround without checking the balance sheet; anchoring on a peak multiple from a different rate regime.
Outcome error — the decision wins in most runs but lost in this one. Examples: a well-researched name gets hit by a sector-wide ETF flow unwind; a hostile regulatory ruling that no analyst flagged; a one-off accounting restatement.

The trap: every loser feels like an outcome error when you're holding the bag. "Nobody could have known." Often someone did know. Short-seller reports, the 10-K risk factors section, a sell-side analyst who'd been at three on a five-point scale — the question is whether you could reasonably have known given how you do research.

A useful test: show your pre-trade notes to a smart friend who doesn't own the stock. Ask them to poke holes. If they find a hole in two minutes that you missed for two months, that's process.

What Good Process Documentation Looks Like

You cannot run a real post-mortem without a real record. The minimum viable version:

Entry note: thesis in 3-5 sentences, the 2-4 pillars it rests on, what would make you exit, position size and why.
Update log: when something material happens — an earnings print, a guide-down, a competitor move — write one or two sentences on whether it confirms or breaks a pillar.
Exit note: what actually triggered the sale, and which pillar (if any) failed.

This takes about 15 minutes per position per quarter. It's the cheapest edge available to a part-time investor.

What to Watch Next

Run post-mortems on your three biggest losers of the past 24 months. Use the five-question framework. Tag each as process or outcome.
Run them on your three biggest winners too. This is the harder discipline — find the trades where you got paid for a bad process. Those are the ones that will hurt you later.
Build a one-page entry-note template and commit to filling it out for every new position over a 1% weight.
Re-check your sizing rule. If your post-mortems reveal that your losses are concentrated in your largest positions, the issue isn't stock selection — it's that you're sizing up on conviction that isn't calibrated.