Why do you only ever find evidence that you were right?
Once you hold a view, your attention quietly reorganises around it. You notice supporting facts, seek supporting sources, and find the disconfirming ones less convincing than a neutral observer would. This is confirmation bias, and its distinctive danger is that it makes bad research feel thorough — you did read a great deal, and all of it agreed with you.
How it looks from the inside
An investor owns a growing subscription business. Over three quarters, gross margin falls from 64% to 61% to 58%. Each fall gets a reading:
- Quarter one: "Investment phase — they said this on the call."
- Quarter two: "Product mix, one-off."
- Quarter three: "Currency."
Every explanation is individually plausible. Two of them may even be true. But notice what never happened: nobody asked whether three consecutive declines are what a deteriorating business looks like from the inside. The bias did not invent facts, and it did not require dishonesty. It just supplied a fresh, reasonable story each time, which is exactly what makes it invisible.
Meanwhile, the reading list shrinks. The bearish analyst is "talking their book." The critical forum thread is "noise." The sources that survive the filter are the ones that already agree — and the resulting confidence is entirely manufactured.
The fix: decide in advance what would prove you wrong
A view that no possible evidence could refute is not a view; it is an attachment. The professional tool is a written thesis with kill criteria — stated before you need them, in numbers, with a deadline:
I own this because subscriber growth stays above 15% and gross margin holds above 60%. If margin closes below 55% for two consecutive quarters, or growth drops below 8%, the thesis is broken and I sell — regardless of the price or the story attached to it.
That sentence does three things at once. It names the evidence, so you cannot quietly change the subject. It sets a threshold, so "a bit weaker" cannot be argued into "fine." And it decides the action while you are calm, so the decision does not have to survive being made in a moment of stress.
Hindsight bias, the sibling
After an outcome is known, it feels as though it was always predictable. The crash was obvious; the winner was clearly a winner. Hindsight bias is comfortable and expensive: if everything looks foreseeable afterwards, you learn nothing about which parts of your process were sound and which merely got lucky.
The antidote is a decision journal — written at the moment of the decision, never after. Record: what you decided, the reasons, what you expected to happen, and how confident you were. Read it a year later. It will disagree with your memory, and the disagreement is the entire value.
A journal also separates the two things beginners conflate: good decisions and good outcomes. A well-reasoned position that lost money and a reckless one that made money are not equal, and only a contemporaneous record can tell them apart.
Try it now
- Pick your highest-conviction holding and write its kill criteria: two numbers and a deadline. If you cannot name evidence that would change your mind, you have found something worth knowing about yourself.
- Read eight quarters of the specific metric you named — revenue growth, margin, or whatever your thesis actually rests on. Every listed company files a statement each quarter, so it is eight rows rather than a memory. Open your holding in the Terminal (the link starts on Apple; change the symbol to yours) and read the quarters off its fundamentals tab. Check whether the trend supports you, or whether you have been supporting it. Apple's newest quarter, so you know the rows:
Open AAPL.US — fundamentals in the EODHD Terminal
- Deliberately read one well-argued piece against your largest position and write the strongest version of its case in your own words. Steel-manning it, not dismissing it, is the exercise.