honesty / graveyard
The graveyard
as of 2026-07-22
Every idea we tested is kept here, including the ones that failed. A 'confirmed' hypothesis held up in its own test; that is not the same as shipping, and none of the 60 candidate signal classes survived the multiplicity correction. Source: evidence packet, section 4 (corrected 2026-09-03).
nulls (351) outnumber confirms (168) 2.1x -- we publish our nulls
nulls by sport (validation ledgers)
| Sport | Ideas tested | Held up in its own test | No effect found | Could not be tested | Did not hold up when re-run |
|---|---|---|---|---|---|
| NBA | 85 | 45 | 31 | 7 | 0 |
| MLB | 86 | 32 | 37 | 8 | 7 |
| Soccer | 64 | 29 | 24 | 8 | 1 |
| Tennis | 52 | 24 | 20 | 8 | 0 |
| Interaction factory (automated pairings) | 1003 | 38 | 239 | 709 | 17 |
reject ledger
804 total recorded verdict rows across all reruns, vs 68 distinct (sport, signal) currently sitting on a REJECT-family verdict as of their latest test -- the gap is signals retested multiple times (asof-reclaim sweeps, re-gating) plus rows whose latest verdict flipped to SHIP.
mechanism survival (testable hypotheses)
Survival rate = share of TESTABLE hypotheses whose latest recorded verdict is CONFIRMED_LOCAL / CONFIRMED_LOCAL_incl_2026_OOS / REPLICATED. A NULL/REJECT is honest market-efficiency evidence, not a failure. No money/advantage/return claim.
verdicts and gate statuses by type (full history)
| REJECT | 627 |
| gateSHIPpassed every gate(95 = SHIP rows the per-signal ledger gate logged across all reruns; none survived the multiplicity correction, so none shipped) | 95 |
| gateSHIP_REVIEWpassed, awaiting human review | 45 |
| NOT_TESTABLE | 14 |
| gateWORSEworse than baseline | 6 |
| INSUFFICIENT | 6 |
| NOT-REPLICATED | 4 |
| gateSTANDING_BASELINE_SETbaseline recorded | 3 |
| DEFER | 1 |
| TIER3_BLOCKED | 1 |
| UNDERPOWERED | 1 |
| gateNULL_UNDERPOWEREDnull at this sample size | 1 |
Gate ledger statuses, not packet verdicts: SHIP = passed every gate; SHIP_REVIEW = passed, awaiting human review; WORSE = worse than baseline; STANDING_BASELINE_SET = baseline recorded; NULL_UNDERPOWERED = null at this sample size.
reject reason categories
survival rate by sport
| Sport | Hypotheses | Could be tested | Held up in its own test | Share that held up |
|---|---|---|---|---|
| NBA | 85 | 78 | 45 | 57.7% |
| MLB | 86 | 78 | 31 | 39.7% |
| Soccer | 64 | 56 | 29 | 51.8% |
| Tennis | 52 | 44 | 24 | 54.5% |