AI · US Tech — Track record
Every elapsed decision is graded against what prices actually did — hit-rates and errors per signal family, plus the cost of the decisions this book chose not to take.
94 graded 0 pending
1W calibration 0.04 How far along the 1-week path a decision typically gets: the fraction of the decided move realized after one week, measured across this book's graded decisions. Feeds a bounded conviction adjustment back into future decisions.
Scorecard
Hit-rates per signal family.
| Signal family | Dir | n | Hit 1W | Hit 4W | Hit 12W | Med err 1W | Stop-first | Bias | Status |
|---|---|---|---|---|---|---|---|---|---|
| FailedBreakdown_Up_20D | sell | 22 | 32% | 50% | — | — | 0% | -4 | ● active |
| NoTrade | sell | 19 | 71% | 73% | — | — | 0% | +4 | ● active |
| FailedBreakout_Down_20D | sell | 9 | 71% | 71% | — | — | 0% | +4 | ● active |
| FailedBreakdown_Up_20D | buy | 8 | 63% | 13% | — | -1.0% | 38% | +2 | ● active |
| RallySell_BearTrend | sell | 7 | 60% | 0% | — | — | 0% | +2 | ● active |
| BreakoutUp_52W | buy | 6 | 50% | 20% | — | -2.9% | 17% | 0 | ● active |
| AnticipationUp_VCP | sell | 4 | 75% | 25% | — | — | 0% | 0 | ○ building sample (4/5) |
| BreakdownDown_20D | sell | 4 | 25% | — | — | — | 0% | 0 | ○ building sample (4/5) |
| BreakoutUp_20D | buy | 4 | 0% | 0% | — | -4.5% | 100% | 0 | ○ building sample (4/5) |
| AnticipationUp_VCP | buy | 2 | 50% | 0% | — | +2.5% | 50% | 0 | ○ building sample (2/5) |
| MomentumContinuation_Up | buy | 2 | 50% | 0% | — | +11.3% | 50% | 0 | ○ building sample (2/5) |
| PullbackBuy_VWAP | buy | 2 | 0% | 50% | — | -1.4% | 100% | 0 | ○ building sample (2/5) |
| PullbackBuy_VWAP | sell | 2 | 50% | 0% | — | — | 0% | 0 | ○ building sample (2/5) |
| BreakoutUp_52W | sell | 1 | 100% | 100% | — | — | 0% | 0 | ○ building sample (1/5) |
| ContinuationLiteUp | sell | 1 | 100% | — | — | — | 0% | 0 | ○ building sample (1/5) |
| ReversalDown_Overbought | sell | 1 | 0% | 100% | — | — | 0% | 0 | ○ building sample (1/5) |
Counterfactuals Counterfactuals grade decisions that were NOT executed — expired queues, no-chase rules, untaken optional trims — against what the market then did. They are kept strictly separate from the calibration cells above.
What the decisions we didn't take would have done.
- Queue expiry (order lapsed unfilled) — 9 cases , median would-have 1W move +12.6% → verdict: cost
- Optional trims not taken — 218 cases , median would-have 1W move +0.8% → verdict: saved
- No-chase queue (queued a pullback instead of chasing) — 14 cases , median would-have 1W move +2.8% → verdict: cost
- Anticipate slot (pre-breakout buy-stop) — 11 cases , median would-have 1W move -2.3% → verdict: saved
Counterfactuals are advisory evidence for the book's learnings reviews — never blended into the calibration cells.