Orca Whirlpool CLMM · 23-pool founding cohort · pre-registered before the
outcome was known · governed by 20260806_220100_solana_acceptance_standard.json
SOL v1 is retired (2026-09-21). No new predictions are being made here. Its successor,
SOL v2, publishes a daily tier for every tracked pool. The 12 v1 cohorts
frozen 09-10 to 09-21 still receive one verdict each as they mature (09-24 to 10-05), exactly as
pre-committed.
This is the graded board: frozen predictions and, once they mature, their outcomes.
Metrics named in the embargo below are withheld deliberately, not missing.
Verdict — evaluation order: INVALID → UNTESTED → EVIDENCE AGAINST → EVIDENCE OF SKILL
UNTESTED
No EVIDENCE AGAINST trigger has fired.
Interval-coverage failure, reported — does not change the tier.
6 of 23 realized_net values fall outside the frozen interval
(2 below · 4 above); P(X>=6|Bin(23,0.10)) = 0.0226.
It is reported alongside the verdict per the standard's one-line commitment, but under
the pre-registered evaluation order the cohort is evaluated as
UNTESTED first (E = 0.2886 < 3.45 and 0 realized traps) — the coverage count
does not change the tier.
Verdict rendered from the signed verdict artifact.
The verdict, E and interval-coverage count on this board are taken from
20260817_140500_sol_verdict_20260813.json
(the committed pre-registration artifact with its own JCS addendum_hash), not
recomputed here. A mutable board must not silently disagree with the committed verdict.
Cohort
23 pools
freeze ts 1785429653
E = Σ y_prob
0.2886
below the 3.45 boundary
Resolved
23 / 23
matures 2026-08-13
Interval misses
6
2 below · 4 above
Realized traps
0
y_true = 1
The band is not a position choice. STATED PLAINLY BECAUSE IT CHANGES HOW EVERY y_prob SHOULD BE READ. frozen_band is the collector's TickArray RPC coverage window, not an LP's chosen range. It is (2N+1) x 88 x tick_spacing ticks, with N from BAND_N_BY_TICK_SPACING (band-v2), centred on whatever tick_current was at genesis 2026-07-16. The collector's own docstring says it 'does NOT gate capture; it is drift datum.' Its width is therefore a function of the pool's Orca fee tier and of an RPC-cost constant - not of any economic judgement about the pool.
Wide-band pools are a volatility forecast. For wide-band pools the position is effectively full range, and 'trap' reduces to a pure price-move test: a full-range CLMM position crosses -5% vs HODL exactly when the 14-day price ratio leaves [0.524, 1.908], i.e. on a 1.91x move in either direction. For those pools y_prob is a volatility forecast and must NOT be described as a pool-quality or yield-trap signal.
Embargoed until the standard's release conditions are met. Withheld:
roc_aucpr_aucbrierprecisionrecalllogistic_recalibrationbootstrap.clusterbootstrap.iid_reference.
Release requires all three: at least 8 resolved NON-OVERLAPPING cohorts; at least 20 realized traps AND at least 20 realized non-traps, pooled; the correlation-adjusted effective sample size reported alongside every figure.
E = Σ y_prob and the interval-coverage count are shown because they are computable from
one cohort and are not on the embargo list.