Open the group chat. Scroll the timeline. Listen to the WhatsApp uncle who's very into commodities now. The phrase shows up everywhere.
I called it.
It comes after the gold rally, after the rate cut, after the IPO that ripped, after the small cap that doubled. Sometimes it's accurate. More often it isn't, and not because the speaker is lying. The brain is a generous editor. By the time the outcome is in the news, it has done quiet work to reshape what was actually said into something that sounds like a clean call.
There are four things a prediction needs in order for "I called it" to mean what it sounds like it means. Most retail predictions, looked at honestly, fail at least three.
1. A time-stamp
A claim made on Tuesday is not the same as a claim made the morning after the print. The first is a prediction. The second is commentary.
This sounds obvious until you try to verify your own. Most people's "calls" live in DMs, voice notes, scattered tweets, conversations at the bar. The artifact of when it was said is whatever your memory provides, which is to say, the flattering version. Without an external time-stamp, every call is retroactively dated to just before the move started. Not because anyone's lying — because that's how memory consolidates an outcome with a prediction. The two get fused into a single narrative, and the narrative is always tidier than the reality.
The fix is mechanical: write it down somewhere that captures the time independently of you. A text to a friend works. A timestamped log works better. A spreadsheet with a date column you can edit doesn't work, even if you swear you'd never edit it.
2. A claim specific enough to be wrong
"I think gold goes higher" is not a prediction. It is a vibe.
A prediction is gold breaks 4,200 by end of Q3. Or Bitcoin trades 90k–110k for the rest of the year. Or Tesla loses its trillion-dollar cap before the next earnings. The reason isn't pedantry — it's that a vague directional claim has no failure mode. Whatever happens, you can find a reading of I think gold goes higher that matches it. If gold goes up 0.3% over six months, you called it. If gold goes up 30% in a month and then crashes back, you called it. If gold trades sideways and you forget about it, you also kind of called it.
The test is whether someone who isn't you, looking at the claim a year later, could decide cleanly whether the prediction came true. If they can't, the prediction wasn't doing the work a prediction is supposed to do. It was decoration.
3. A scenario where you would have been wrong
This is the one almost everyone skips.
A real prediction implies an alternative reality where you'd have to admit the call was off. Gold breaks 4,200 by end of Q3 implies if it's still under 4,200 on October 1, I was wrong. You don't need to write down the alternative explicitly, but it has to be implicit in the shape of the claim. If there's no path through the next year where you'd say that one didn't pan out, the claim wasn't a forecast. It was insurance against future feeling-bad.
The honest test, applied at the moment of the call: can I describe the world in which this turns out to be a bad call? If the answer is no, "I called it" was already pre-written. The market was just a prop.
4. Stakes — but not necessarily money
This one is more subtle. The stakes don't have to be a position. They have to be public enough, or recorded enough, that the prediction exists in the world independent of how it turned out.
The reason is that hindsight bias does its damage between the prediction and the outcome. If the prediction lives only in your head, the brain is free to weight it later — to remember the call vividly when it pays off and forget it cleanly when it doesn't. Most retail traders genuinely believe they have a better track record than they do, not because they're dishonest, but because the survivors of their predictions are the only ones still in the dataset their memory carries.
Stakes break the loop. A prediction told to a friend who'll bring it up later. A logged hunch you have to look at next quarter. A bet for a coffee. Anything that makes the prediction outlive its outcome in a form you can't edit. That's the discipline. Money on the trade is one version of stakes; it isn't the only version, and it isn't always the cleanest.
What survives the four
Apply those four to the calls you've made over the last year. Most won't survive. The honest count of real predictions you made and got right is almost always lower than the count of times you've thought "I called it." That isn't a character flaw. It's the default state of an unaudited memory operating on outcomes that are loud and predictions that are quiet.
You can't fix this by trying harder to remember. The fix is structural: a record outside your head, made before the answer is known, specific enough to be wrong, and durable enough that you have to reckon with it later.
That's all HunchLog is, in the end. A record outside your head. The discipline is in the design — not in remembering to be honest.
If you said I called it this year, the question worth asking isn't was I right. It's would the four things above hold up under audit. If they would, that prediction is one of the calls that compounds into self-knowledge. If they wouldn't, you got lucky on a vibe — and the next time the vibe doesn't pay, the same brain will quietly forget that one too.