Predictions
Dated calls, resolved in public — including the misses. A prediction without a date and a resolution is a mood.
Undated forecasts cost nothing.
Most industry prediction is written to be un-losable: no date, no threshold, no way to be wrong. That is commentary, not judgement, and it never has to pay for itself.
Every call here carries the date it was made, the condition that would resolve it, and the resolution when it arrives.
A call is only a call if it can fail.
The misses stay up.
A prediction log that only shows the hits is a portfolio page. Ours keeps everything, and the miss notes are usually more useful than the hits — they show where the model of the market was wrong, not just where it was lucky.
We change our minds in public. That is one of the six published principles, and this is where it costs us something.
Where we actually have a position.
Three predictions, on the record.
Published alongside the log rather than claimed here. The number matters less than whether the reasoning behind the misses was recoverable.
Yes. We only publish predictions we would act on ourselves, so not every suggestion becomes one.
Never the call. Resolution notes are appended with their own dates.
We change our minds in public.
Including when wrong — that is the point of publishing the date.
Read the Journal