$ calibration --by-conviction
Does your conviction predict your returns?
For every buy, the conviction you had recorded at the time, against the return from the price you paid to the most recent close. It measures the scores you acted on — conviction only costs you anything when you put money behind it.
Your lower-conviction buys have done better — 5–6 leads 9–10 by 45.8 points. Worth asking what the confident ones had in common.
return by conviction at time of purchase
- 1–4doubtfulno buys
- 5–6on the fence+22.71%6 buys · $61,959.25 committed · plain mean 18.46%
- 7–8confident-6.26%4 buys · $81,712.58 committed · plain mean -7.95%
- 9–10all in-23.14%3 buys · $43,991.73 committed · plain mean -23.06%
Percentages are weighted by money committed. The plain mean is shown too — one large trade can carry the weighted figure on its own, and the two disagreeing is itself worth noticing.
take this to an LLM
Copies the record above with instructions to look for what your confident mistakes have in common — and to say so if the sample is too thin to tell.
Numbers only. The written theses stay on their own pages, where each position has its own copy button.