Daily registry · Oct 5, 6:20 PM ET
v5.1 seat registry
Snapshot for 2026-10-05, built 2026-10-05 22:20 UTC on Nicholass-iMac.local. Daily report on every Table 1 and Table 2 rule by sport. Every number is read from the sport's own file on origin/main; a dash means the sport does not record it.
- 16 Table 1 seats
- 14 live
- 2 waiting
- 32 paper
Table 1: soccer 6 (4 live) · MMA 1 live · NFL 2 live · CFB 1 live · NHL 1 live · NBA 4 live · MLB 1 live · paper 32 · changes: NFL rule dropped Totals UNDER: no_extra_rest + home_last_under; NFL rule dropped home_hc12 BET; NFL rule dropped away_dproe_lo BET; +1 more
- Soccer T1 4/6 live · T2 7 · unverified 3
- MMA T1 1/1 live · T2 2 · unverified 3
- NFL T1 2/2 live · T2 6
- College football T1 1/1 live · T2 3
- NHL T1 1/1 live · T2 2
- NBA T1 4/4 live · T2 7
- MLB T1 1/1 live · T2 3
- Tennis T1 0/0 live · T2 2
✓ passes ✗ misses … blind still accruing, not losing ? the sport does not record it ⚠ the sport's verdict and its values disagree One mark per v5.1 criterion (guide SECTION 1): proof, seasons, CLV, blind holdout, ROI band, integrity. A miss shows the grader's reason under the row.
Changed since 2026-10-04
- NFL rule dropped Totals UNDER: no_extra_rest + home_last_under — was paper
- NFL rule dropped home_hc12 BET — was paper
- NFL rule dropped away_dproe_lo BET — was paper
- NFL rule dropped home_proe_lo BET — was paper
Soccer
Closest to a seat: 6 seats live (4 posting); 4 one-miss rules with >= 150 winning blind bets (see soccer table)
Biggest obstacle: Owner rule: must win on Kalshi AND Polymarket
Top leads
Ten-league M005/M009/M010 forward
Read from: SOCCER AI/registries/systems_registry.json @ origin/main 2706edab (2026-10-03) · SOCCER AI/registries/table1_live.json @ origin/main af4cb4fa (2026-10-01) · SOCCER AI/workers/picks-cron/src/v51Gates.js @ origin/main 123ec14f (2026-10-02) · SOCCER AI/data/more_leagues_forward/LEDGER.json @ origin/main 49bd1e40 (2026-10-03)
10 of 10 registered paper rules accounted for (7 listed, 3 unverified) (counted in SOCCER AI/registries/systems_registry.json).
Table 1 live (4)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| Shallow Gap Lucky Draw | 265 | 29.0% | 0.012 | 5.8% | 5/5 | 434 · 0.93% · 0.006 | 169/150 · full blind seasons 0/2 (≥ 75) · 3.9% | passed (one flag) | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Sealed 74-191-0 (+76.9u) · Blind 39-130-0 (+6.6u) | |||||||||
| Close Ratings Host Draw | 184 | 31.9% | 0.005 | 9.1% | 3/3 | 439 · 1.07% · <0.001 | 255/150 · full blind seasons 0/2 (≥ 75) · 12.7% | passed (one flag) | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 65-119-0 (+58.6u) · Blind 80-175-0 (+32.3u) | |||||||||
| Low Corner Pair Draw | 204 | 22.6% | 0.022 | 2.8% | 3/3 | 473 · 1.26% · <0.001 | 269/150 · full blind seasons 0/2 (≥ 75) · 10.4% | passed (one flag) | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 71-133-0 (+46.1u) · Blind 84-185-0 (+28.0u) | |||||||||
| Rated Host Low Corners | 214 | 20.5% | 0.035 | 1.3% | 2/3 | 511 · 1.85% · <0.001 | 297/150 · full blind seasons 0/2 (≥ 75) · 15.1% | passed (one flag) | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 72-142-0 (+43.9u) · Blind 89-208-0 (+44.8u) | |||||||||
Table 1 approved, waiting to go live (2)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Missing | |
|---|---|---|---|---|---|---|---|---|---|
| Low Corner Model Draw | 266 | 24.1% | 0.004 | 8.3% | 3/3 | 614 · 1.14% · <0.001 | 348/150 · full blind seasons 0/2 (≥ 75) · 2.9% | decides on a Pinnacle-open price; v51Gates.js PRICE_TWIN_LIVE is false until the twin capture is read back | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 103-163-0 (+64.2u) · Blind 113-235-0 (+10.2u) | |||||||||
| Shallow Box Priced Draw | 191 | 19.9% | 0.029 | 1.9% | 3/3 | 498 · 1.09% · <0.001 | 307/150 · full blind seasons 0/2 (≥ 75) · 11.5% | decides on a Pinnacle-open price; v51Gates.js PRICE_TWIN_LIVE is false until the twin capture is read back | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 74-117-0 (+38.0u) · Blind 111-196-0 (+35.2u) | |||||||||
Table 2 paper (7)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| Lucky Host Toothless Draw | 220 | 16.4% | 0.087 | -4.4% | 4/5 | 391 · 0.93% · <0.001 | 171/150 · full blind seasons 0/2 (≥ 75) · 6.1% | passed (one flag) | no forward ledger is defined for this sport |
✗ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity
Sealed 64-156-0 (+36.2u) · Blind 46-125-0 (+10.5u) | |||||||||
| English Slump Away Value | 197 | 55.3% | <0.001 | 21.6% | 4/4 | 368 · 1.40% · 0.026 | 171/150 · full blind seasons 0/2 (≥ 75) · 24.7% | passed (one flag) | no forward ledger is defined for this sport |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✗ ROI band ✓ Integrity
Sealed 65-132-0 (+109.0u) · Blind 34-137-0 (+42.2u) | |||||||||
| Shallower Box Blunt Draw | 299 | 19.0% | 0.016 | 4.0% | 3/5 | 575 · 1.43% · <0.001 | 276/150 · full blind seasons 0/2 (≥ 75) · 1.3% | passed (one flag) | no forward ledger is defined for this sport |
✓ Proof ✗ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity
Sealed 109-190-0 (+56.9u) · Blind 83-193-0 (+3.5u) | |||||||||
| Mild Luck Passive Draw | 303 | 22.7% | 0.013 | 3.7% | 3/5 | 464 · 0.76% · 0.003 | 161/150 · full blind seasons 0/2 (≥ 75) · 3.7% | passed (one flag) | no forward ledger is defined for this sport |
✓ Proof ✗ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity
Sealed 90-213-0 (+68.7u) · Blind 43-118-0 (+5.9u) | |||||||||
| Home Edge Low Corner Draw (Ten Leagues) | 402 | 16.3% | 0.022 | 2.6% | 3/3 | 1167 · 1.44% · <0.001 | 765/150 · full blind seasons 0/2 (≥ 75) · 1.7% | passed (one flag) | 19 stamped rows, file 2026-10-05 15:02 UTC SOCCER AI/data/more_leagues_forward/stamps.jsonl (all three ten-league rules write one file; stamped counts the rows where this rule fired) |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 134-268-0 (+65.5u) · Blind 218-547-0 (+12.6u) forward: 0 stamped / 150 · frozen 2026-10-02 | |||||||||
| Elo Parity Low Corner Draw (Ten Leagues) | 1425 | 7.2% | 0.036 | 0.5% | 3/3 | 3526 · 1.06% · <0.001 | 2101/150 · full blind seasons 0/2 (≥ 75) · 1.4% | passed (one flag) | 19 stamped rows, file 2026-10-05 15:02 UTC SOCCER AI/data/more_leagues_forward/stamps.jsonl (all three ten-league rules write one file; stamped counts the rows where this rule fired) |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 469-956-0 (+102.8u) · Blind 643-1458-0 (+29.8u) forward: 3 stamped / 150 · frozen 2026-10-02 | |||||||||
| Away Drawer Rest Edge Draw (Ten Leagues) | 267 | 17.2% | 0.039 | 0.2% | 3/3 | 605 · 0.64% · 0.003 | 338/150 · full blind seasons 0/2 (≥ 75) · 0.2% | passed (one flag) | 19 stamped rows, file 2026-10-05 15:02 UTC SOCCER AI/data/more_leagues_forward/stamps.jsonl (all three ten-league rules write one file; stamped counts the rows where this rule fired) |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Blind, ROI band. Every other mark is the sport's own verdict. Sealed 89-178-0 (+45.9u) · Blind 95-243-0 (+0.8u) forward: 1 stamped / 150 · frozen 2026-10-02 | |||||||||
Unverified (3)
Guide v3.5.0, verified means recorded: these rules are listed so nobody counts them. Not in Table 1, Table 2 or the paper count; never posted.
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Why unverified | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| Strong Luck Passive Draw | 161 | 21.1% | 0.071 | -5.9% | 4/5 | 240 · 0.73% · 0.025 | 79/150 · full blind seasons 0/2 (≥ 75) · 20.5% | Table 2: no per-rule forward log: its blind bets cannot be counted to 150 | no forward ledger is defined for this sport |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 46-115-0 (+34.0u) · Blind 25-54-0 (+16.2u) | |||||||||
| H-LOOP-114 | 300 | 26.4% | 0.011 | 5.3% | 7/8 | 300 · -0.53% · 0.779 | 0/150 · full blind seasons 0/2 (≥ 75) · — | Table 2: CLV miss with the wrong sign: mean CLV -0.53% is not > 0 (a fail, not a near-miss) no per-rule forward log: its blind bets cannot be counted to 150 | no forward ledger is defined for this sport |
✓ Proof ✓ Seasons ✗ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 86-214-0 (+79.1u) · Blind 0-0-0 (+0.0u) | |||||||||
| H-LOOP-132 | 131 | 30.8% | 0.002 | 11.5% | 3/4 | 131 · 1.94% · 0.002 | 0/150 · full blind seasons 0/2 (≥ 75) · — | Table 2: no per-rule forward log: its blind bets cannot be counted to 150 | no forward ledger is defined for this sport |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 70-61-0 (+40.4u) · Blind 0-0-0 (+0.0u) | |||||||||
MMA
Closest to a seat: **Statistical Main Card Experienced is SEATED (v5.1, Kalshi tickets)**; integrity written down 2026-10-03 (prices 1017/1017, look-ahead 2034/2034, engine B agrees). **Live lane fixed 2026-10-04 (mma-lane-origins, mmalogic PR #61-63, hub #39/#41): 10-10 dry run 2 fires, both origins agree; /v51 live true; page note: blind 2015-19 is a backcast (constants tuned 2020-26). First live card Sat 2026-10-10 (Allen vs Duncan). Killed 2026-10-04 (mmalogic 758735ae): early drift (0/4 cells, -5.7% to -9.5% at the consensus open) and last-fight shape (0/10; best fade-off-KO-loss -1.27%). Fresh hunt 2026-10-04 (mmalogic c7ea83d1): nationality family rejected (fade_us dies blind); home_abroad (home fighter at a non-US card) listed Table 2 paper (proof p .0527 only; blind 0/150 forward from the 2026-10-04 freeze; leaked 2012-20 +9.61% n699 disclosed, never counted) — /v51 reads it clean 01:30Z after the iMac mmalogic checkout was fast-forwarded to c7ea83d1. Next untested: model-agnostic age gap (fightodds birthDate; blind 2012-14 only).** Next: 3 Table 2 tracking rows (below) each need a two-engine replay
Biggest obstacle: Only 1 of 42 graded systems passes; the other 38 miss proof and holdout at the real open. Two-engine replay (`scripts/v51_engine_b.py`) is written for the seated filter only
Top leads
Statistical Body Leg Experienced (misses only holdout: no 2015-19 blind bets possible), H_REACH_ACC_ADV_ML (misses proof at p .0517, n 97; blind 74 at +23.52%), Value Decision Favorite (misses proof at p .0601)
Read from: mmalogic/docs/reports/system_performance_registry.json @ origin/main 25873db9 (2026-10-03) · mmalogic/docs/reports/v51_seats.json @ origin/main 427e7f44 (2026-10-03) · mmalogic/docs/reports/v51_prehistory_look_2026-10-02_frozen.json @ origin/main 461b1bbe (2026-10-02) · mmalogic/docs/reports/v51_table2.json @ origin/main 2ffd5ce3 (2026-10-04) · mmalogic/docs/reports/v51_prehistory_look_2026-10-03_regrade_v33.json @ origin/main c76a4ff5 (2026-10-03) · mmalogic/docs/reports/v51_look_home_abroad_2026-10-04.json @ origin/main 05ecbf80 (2026-10-04) · mmalogic/docs/reports/v51_look_young8_2026-10-04.json @ origin/main 2ffd5ce3 (2026-10-04)
Statistical Main Card Experienced: last live-lane pass 2026-10-03T19:56 (event None): refused (live tickets start with the card dated 2026-10-04; this card is 2026-10-03 (dry run only))
Statistical Main Card Experienced: blind 2015-19 is a backcast: constants tuned on 2020-26 (orchestrator ruling 2026-10-04: the owner-approved past-window method, one look at a held-out earlier window).
MMA has no paper betting lane (mmalogic CLAUDE.md); a seat is read from docs/reports/v51_seats.json, and Table 2 rows are tracking listings from docs/reports/v51_table2.json (no bets, no tickets).
5 of 5 registered paper rules accounted for (2 listed, 3 unverified) (counted in mmalogic docs/reports/system_performance_registry.json + docs/reports/v51_table2.json (tracking listings) + docs/reports/v51_seats.json).
Table 1 live (1)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| Statistical Main Card Experienced | 597 | 7.2% | 0.023 | 1.3% | 5/7 | 1017 · 1.77% · <0.001 | 420/150 · full blind seasons 0/2 (≥ 75) · 4.6% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Sealed 383-214-0 (+43.1u) (an earlier book, mmalogic/docs/reports/v51_open_price_bets.json, holds 588 bets) · Blind 264-156-0 (+19.3u) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (2)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| home_abroad | 150 | 12.2% | 0.053 | 0.2% | 3/4 | 150 · 1.80% · 0.025 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 1 stamped rows, file 2026-10-05 01:30 UTC mmalogic/data/forward/ufc_v51/table2/home_abroad.jsonl |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Record (sealed and blind): not recorded (mmalogic/docs/reports/v51_look_home_abroad_2026-10-04.json — no graded rows for this listing) forward: 0 stamped / 150 · frozen 2026-10-04 | |||||||||
| young8 | 261 | 6.9% | 0.079 | -1.0% | 5/6 | 261 · 4.46% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 4 stamped rows, file 2026-10-05 01:52 UTC mmalogic/data/forward/ufc_v51/table2/young8.jsonl |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Record (sealed and blind): not recorded (mmalogic/docs/reports/v51_look_young8_2026-10-04.json — no graded rows for this listing) forward: 3 stamped / 150 · frozen 2026-10-04 | |||||||||
Unverified (3)
Guide v3.5.0, verified means recorded: these rules are listed so nobody counts them. Not in Table 1, Table 2 or the paper count; never posted.
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Why unverified | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| H_REACH_ACC_ADV_ML | 97 | 14.0% | 0.052 | 0.5% | 3/4 | 171 · 3.35% · <0.001 | 74/150 · full blind seasons 0/2 (≥ 75) · 23.5% | Table 2: Integrity not proven: sport's integrity flag is NOT proven no per-rule forward log: its blind bets cannot be counted to 150 | — |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✗ Integrity
Sealed 68-29-0 (+13.6u) · Blind 50-24-0 (+17.4u) | |||||||||
| Statistical Body Leg Experienced | 467 | 8.0% | 0.028 | 1.3% | 5/6 | 467 · 2.24% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | Table 2: Integrity not proven: sport's integrity flag is NOT proven no per-rule forward log: its blind bets cannot be counted to 150 | — |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✗ Integrity
Sealed 298-169-0 (+37.3u) · Blind 0-0-0 (+0.0u) | |||||||||
| Value Decision Favorite | 151 | 18.1% | 0.060 | -1.8% | 3/3 | 151 · 1.29% · 0.032 | 0/150 · full blind seasons 0/2 (≥ 75) · — | Table 2: Integrity not proven: sport's integrity flag is NOT proven no per-rule forward log: its blind bets cannot be counted to 150 | — |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✗ Integrity
Sealed 62-89-0 (+27.4u) · Blind 0-0-0 (+0.0u) | |||||||||
NFL
Closest to a seat: Method 2 on the vocabularies EXP-110 left out killed 2026-10-05 (EXP-114: 2 totals UNDER passed sealed, both lost blind 2007-13); 2007-13 is now spent for every frozen NFL vocabulary; FWD-005, home_hc12, away_dproe_lo, home_proe_lo are forward-only (lost their one blind look). New NFL families are forward-only. Implied-team-total censoring killed on sealed 2014-25 2026-10-05 (EXP-113, Football 8c317b3; 2007-13 still unread by it). Next: /fresh-sports-idea-hunt nfl with a 2007-13 lineage grep first; or method 2 (re-mine NFL vocabularies NOT in EXP-110 rounds 3-6, blind 2007-13). Totals UNDER (no extra rest + home last under): passes v5.1, owner chose practice 2026-10-03; forward stamps from week 5. Next: `away_dproe_lo` BET (p .018, one-miss on book CLV p .10 — the nearest NFL miss yet), then `home_hc12` BET and `home_proe_lo` BET (CLV only); each needs venue CLV on >= 100 forward bets
Biggest obstacle: No unread past window left (2014-22: spreads 5 looks, totals 1 + 9 reads); 150 blind bets per PROE rule = 2.1-2.3 seasons (2028); no free real opening juice before 2014
Top leads
6 forward paper rules (blind 0/150); totals rule forward; `home_hc12` BET, `away_dproe_lo` BET and `home_proe_lo` BET forward from week 5 (FWD-006, FWD-007); tempo / pace (EXP-107) and PROE (EXP-108) read on 2023-25, 0 passers; every further family is forward-only
Read from: Football/data/registry/nfl_systems_registry.json @ origin/main bbb53c47 (2026-10-04) · Football/data/backtest_artifacts/v51/open_replay.json @ origin/main adb7c9e0 (2026-10-01) · Football/data/backtest_artifacts/v51/restate_2026-10-02_paper_rules.json @ origin/main b272ff4e (2026-10-02) · Football/docs/preregistration/nfl-v51-totals-v1.frozen.json @ origin/main 9c38183a (2026-10-02) · Football/docs/preregistration/nfl-v51-round4-families-v1.forward.json @ origin/main ffe45236 (2026-10-03) · Football/docs/preregistration/nfl-v51-round6-families-v1.forward.json @ origin/main 161c5de7 (2026-10-03) · Football/docs/preregistration/nfl-v51-totals-v1.forward.json @ origin/main 46f2ed06 (2026-10-03) · Football/data/backtest_artifacts/v51/gate_values_2026-10-03.json @ origin/main 0b1d1c0c (2026-10-03) · Football/docs/preregistration/nfl-v51-remine2-v1.blind.json @ origin/main b4483c3b (2026-10-05) · Football/docs/preregistration/nfl-v51-remine-v1.blind.json @ origin/main a6ecb0b6 (2026-10-03)
Forward paper stamps (one file per rule under data/forward) are written by the iMac job (first spread-rule run Mon 2026-10-05) and are not all in git; blind counts are the registry's.
Away Rebuilder Passing Edge shows its restate (151 sealed bets; EXP-103, restate_2026-10-02_paper_rules.json); the original (125 bets) is in open_replay.json.
Totals UNDER: no_extra_rest + home_last_under is forward-only, not Table 2 (orchestrator 2026-10-05): sealed 2014-25 7/12 seasons profitable; blind 2007-13 at the proxy open -5.60% on 532 bets (EXP-114, Football docs/preregistration/nfl-v51-remine2-v1.blind.json). Its practice stamps continue.
home_hc12 BET is forward-only, not Table 2 (orchestrator 2026-10-05): sealed 2014-25 3/12 seasons profitable; blind 2007-13 at the proxy open -5.35% on 613 bets (EXP-110, Football docs/preregistration/nfl-v51-remine-v1.blind.json). Its practice stamps continue.
away_dproe_lo BET is forward-only, not Table 2 (orchestrator 2026-10-05): sealed 2014-25 4/12 seasons profitable; blind 2007-13 at the proxy open -15.82% on 381 bets (EXP-110, Football docs/preregistration/nfl-v51-remine-v1.blind.json). Its practice stamps continue.
home_proe_lo BET is forward-only, not Table 2 (orchestrator 2026-10-05): sealed 2014-25 3/12 seasons profitable; blind 2007-13 at the proxy open -6.39% on 382 bets (EXP-110, Football docs/preregistration/nfl-v51-remine-v1.blind.json). Its practice stamps continue.
6 of 6 registered paper rules listed (counted in Football data/registry/nfl_systems_registry.json + docs/preregistration/nfl-v51-totals-v1.frozen.json + nfl-v51-round4-families-v1.forward.json + nfl-v51-round6-families-v1.forward.json + nfl-v51-remine2-v1.blind.json + nfl-v51-remine-v1.blind.json).
Table 1 live (2)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| home_dproe_hi UNDER | 687 | 8.5% | 0.010 | 2.6% | 9/12 | 1142 · 1.80% · <0.001 | 455/150 · full blind seasons 0/2 (≥ 75) · 3.7% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2007-13, guide v3.6.0); the whole search would produce a false Table 1 candidate 57% of the time with no edge — real-price results are tracked from week 5. Sealed 382-293-12 (+58.2u) · Blind 236-209-10 (+16.7u) | |||||||||
| home_lowbrand FADE | 471 | 8.1% | 0.031 | 1.0% | 8/12 | 899 · 0.54% · 0.049 | 428/150 · full blind seasons 0/2 (≥ 75) · 6.9% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2007-13, guide v3.6.0); the whole search would produce a false Table 1 candidate 57% of the time with no edge — real-price results are tracked from week 5. Sealed 255-197-19 (+38.1u) · Blind 224-188-16 (+29.6u) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (6)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| Away Rebuilder Passing Edge | 151 | 15.0% | 0.026 | 2.6% | 8/10 | 151 · 3.62% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-05) Football/data/forward/nfl_de_open_v1/v51_paper/away-rebuilder-passing-edge/stamps.jsonl — football.forward-open first run Mon 2026-10-05 12:00 CT (tasks/todo.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 87-58-6 (+22.6u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-01 (first run pending 2026-10-05) | |||||||||
| Home Rebuild | 100 | 28.9% | 0.001 | 14.5% | 3/3 | 100 · 1.52% · 0.038 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 4 stamped rows, file 2026-10-05 17:48 UTC Football/data/forward/nfl_de_open_v1/v51_paper/home-rebuild/stamps.jsonl |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 66-31-3 (+28.9u) · Blind 0-0-0 (+0.0u) forward: 4 stamped / 150 · frozen 2026-10-02 | |||||||||
| Home Late Covered Season | 101 | 20.0% | 0.017 | 5.0% | 3/3 | 101 · 3.62% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-05) Football/data/forward/nfl_de_open_v1/v51_paper/home-late-covered-season/stamps.jsonl — football.forward-open first run Mon 2026-10-05 12:00 CT (tasks/todo.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 63-37-1 (+20.2u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-05) | |||||||||
| Home Third | 150 | 14.7% | 0.029 | 2.2% | 3/3 | 150 · 2.00% · 0.018 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 4 stamped rows, file 2026-10-05 17:48 UTC Football/data/forward/nfl_de_open_v1/v51_paper/home-third/stamps.jsonl |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 89-59-2 (+22.1u) · Blind 0-0-0 (+0.0u) forward: 3 stamped / 150 · frozen 2026-10-02 | |||||||||
| Away SRS | 120 | 15.7% | 0.034 | 1.9% | 2/3 | 120 · 1.45% · 0.049 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-05) Football/data/forward/nfl_de_open_v1/v51_paper/away-srs/stamps.jsonl — football.forward-open first run Mon 2026-10-05 12:00 CT (tasks/todo.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 71-46-3 (+18.9u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-05) | |||||||||
| Late Season | 191 | 12.0% | 0.040 | 0.9% | 3/3 | 191 · 1.70% · 0.015 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-05) Football/data/forward/nfl_de_open_v1/v51_paper/late-season/stamps.jsonl — football.forward-open first run Mon 2026-10-05 12:00 CT (tasks/todo.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 111-78-2 (+23.0u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-05) | |||||||||
College football
Closest to a seat: **1 Table 1 seat live: CFB-REMINE-03 (criterion 2 corrected to 6/6 sealed, /v51 20:55Z 2026-10-04).** CFB-V51-16: blind 37/150 at +5.87%. Next (cfb-ot-hangover close 2026-10-04, plan docs/plans/2026-10-04_cfb-v51-pcover-2011-13-plan.md): one past-window look at 2011-13 for the 14 pcover/edge paper rules (miss only holdout; ~380-620 projected blind bets each), pcover retrained walk-forward with the frozen code (ruling C). Killed 2026-10-04: overtime hangover (EXP-V51-31, 0/4), comeback / blown lead (EXP-V51-32, 0/4), and the 13 pcover paper rules (same-season calibration look-ahead; restate fails criterion 2; 2011-13 build dead on parity). H-LOOP-017 restored to paper (its HFA path never ran). Owner-supplied YoyoYoshi source EXHAUSTED 2026-10-04 (cfb 1e3ef43): realignment year-1 0/4, familiarity UNDER 1/4 (CLV +2.85% but ROI fails), recruiting vs baseline 0/4 + 1/4; blind never opened. Familiarity ATS killed 2026-10-04 (EXP-V51-37, home -1.16% 1/4, road -7.35%; cfb 6810554). Winless weeks 3-6 killed 2026-10-04 (EXP-V51-38, 0/4 all cells, ~vig; cfb eaf1594). Fresh hunt 2026-10-05 (cfb 3df6557): PGWE luck 0/24; prior gap (fade the team outplaying its prices, weeks 3-7) passed 3/4 sealed (+8.92% n 311, CLV miss) then its one 2011-13 look lost (-6.39% n 230) — dead; 2011-13 spent for the prior-gap atom in any market. REMINE-2 2026-10-05 (cfb f3de174): CFB-REMINE2-01 (opp_bowl_dead & opp_race_out & pick_lookback, sealed 262 +11.00%) lost its one 2011-13 look (-2.32% n 161) — dead; totals re-mine 0/96 pre-screen. Method 2 is out of vocabularies that reach 2011-13 (free play-by-play starts 2014). Owner sources 2026-10-05: fourth-data.com (stakes / off-FCS atoms EXP-V51-42, 0/8) and basedonactualmath.com (only A1 playoff leverage, 100% inside killed title_live — not run); censoring EXP-V51-43 dead (-4.3% / -4.6%). 2011-13 still unread for most atoms; any rule using opp_race_out may not use it. Next: /fresh-sports-idea-hunt cfb or the owner's next source.
Biggest obstacle: Real opening prices only 2014-16 and 2023+ (none 2017-22); free data only
Top leads
16 paper rules accruing (CFB-V51R7-01 added 2026-10-03: dog whose 3rd-down conversion beats its success rate, FORWARD paper only, one-miss on MINED CLV p .435, 2014-16 never its blind window; rounds 5-11 killed (round 9 passing-down 0/71; round 10 clock management untested — free PBP timeouts not on one scale across seasons; round 11 returning production: home dog bottom-20% one-miss on CLV killed by its null, 601/1,000 no-signal worlds reach the pre-screen, values not an archived preseason snapshot; cfb-algorithm 6e56895); next untested: round 12 scripted opening (first-15-play success beyond overall success); CFB-V51R3-01 added 2026-10-03: turnover luck x finishing, one-miss on blind CLV p .065; CFB-V51R4-01 added 2026-10-03: away dog vs a poor kickoff-return team, one-miss on blind CLV p .189, blind edge decays to -1.2% by 2016; forward stamps from 2026-10-03; 2014-16 has had 2 blind looks, 0 seats); totals UNDER + away 2+ time zones (171 / +13.7%, CLV verified on 81 < 100); 2026-12-07 re-run with 2026 added
Read from: cfb-algorithm/data/registry.json @ origin/main 0f3f5a09 (2026-10-04)
Forward paper ledger (data/v51_paper_ledger.db, open stamps per rule) is gitignored and lives on the iMac; blind counts are the registry's.
CFB-V51R4-01 shows its v51_restated_open_exception restate (188 sealed bets); the registered frame (190 bets) is history in data/registry.json.
CFB-V51R7-01 shows its v51_restated_open_exception restate (195 sealed bets); the registered frame (197 bets) is history in data/registry.json.
3 of 3 registered paper rules listed (counted in cfb-algorithm/data/registry.json).
Table 1 live (1)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| CFB-REMINE-03 | 240 | 15.1% | 0.007 | 5.2% | 6/6 | 431 · 1.94% · <0.001 | 191/150 · full blind seasons 0/2 (≥ 75) · 0.9% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on an assumed-price blind window (CFB 2017-22, guide v3.10.0, -110 cut 0.5%): blind +0.86% on 191 bets, and only 1 of the 7 related rules held up blind (family mean blind about -4%) — treat as weak evidence; real-price results are tracked from its first fire (~week 11). Sealed 142-94-4 (+36.2u) · Blind 98-87-6 (+1.6u) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (3)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| CFB-V51R3-01 | 129 | 14.8% | 0.038 | 1.3% | 3/3 | 291 · 1.10% · 0.065 | 162/150 · full blind seasons 0/2 (≥ 75) · 3.6% | origins yes · engines yes | 11 stamped rows, file 2026-10-05 16:09 UTC cfb-algorithm/data/v51_paper_ledger.db:open_stamps |
✓ Proof ✓ Seasons ✗ CLV ✓ Blind ✓ ROI band ✓ Integrity
Sealed 77-51-1 (+19.1u) · Blind 86-72-4 (+5.8u) forward: 11 stamped / 150 · frozen 2026-10-03 | |||||||||
| CFB-V51R4-01 | 188 | 14.8% | 0.016 | 3.6% | —/— | — · 0.36% · 0.251 | 309/150 · full blind seasons 0/2 (≥ 75) · 5.0% | origins yes · engines yes | 19 stamped rows, file 2026-10-05 16:09 UTC cfb-algorithm/data/v51_paper_ledger.db:open_stamps |
✓ Proof ✓ Seasons ✗ CLV ✓ Blind ✓ ROI band ✓ Integrity
Sealed 113-75-0 (+27.9u) · Blind 168-138-3 (+15.3u) forward: 19 stamped / 150 · frozen 2026-10-03 | |||||||||
| CFB-V51R7-01 | 195 | 15.8% | 0.010 | 4.8% | —/— | — · 0.13% · 0.436 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 14 stamped rows, file 2026-10-05 16:09 UTC cfb-algorithm/data/v51_paper_ledger.db:open_stamps |
✓ Proof ✓ Seasons ✗ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 117-76-2 (+30.8u) · Blind 0-0-0 (+0.0u) forward: 14 stamped / 150 · frozen 2026-10-03 | |||||||||
NHL
Closest to a seat: **1 Table 1 seat live: Fade Overtime Hangover.** Killed 2026-10-04 (icebreaker-ai 0ce40c9): last-game shape beyond OT, moneyline 0/84 and totals 0/36 (OT atoms excluded from 2008-19). SBR 2014-19 NHL opening totals are not real opens -> totals blind = 2008-10 only. Coach atoms on totals killed 2026-10-04 (0/20, icebreaker-ai 66cdfcd; coach feed left-censoring fixed). Schedule atoms on totals stopped at lineage 2026-10-04 (a859ba2): B2B / 3-in-4 / rest / points% gap / road-trip length all screened on 2008-20 moneyline, so no honest blind window (2007-08 no totals price, 2014-19 opens fake, 2025-26 no second totals origin). Past-provable NHL levers in current data: none named — next NHL session starts with /fresh-sports-idea-hunt nhl for an atom no 2008-10 family read. RSF + RPF (one family, 225 / 107 sealed, +5.87% / +5.48%, miss only proof p .157 / .257)
Biggest obstacle: 2008-20 is spent for every current idea; no free second close; T_HC_NEW needs positive Kalshi or Polymarket venue CLV on >= 100 forward bets
Top leads
T_HC_NEW and XUB are **forward-only, not Table 2** (T_HC_NEW: CLV -0.47% is wrong-signed; XUB: misses proof and CLV once the spent 2008-20 bets are out); **RLMDOG-NHL** (RLM underdog at T-2h, venue fire-time price) is forward-only, not Table 2: Polymarket fails CLV + seasons + blind, Kalshi fails proof + CLV (98 of 103 carry a close) + seasons + blind; fire stamps live from 2026-10-03; earliest Table 1 ~Jan 2028 (150 blind bets at ~100 a season)
Read from: icebreaker-ai/nhl_system_registry.json @ origin/main f12fb9bf (2026-10-04) · icebreaker-ai/research/v51/grade_forward_frozen.py @ origin/main f6098309 (2026-10-03) · icebreaker-ai/research/v51/results/lever1_v51_v2.json @ origin/main df0485ab (2026-10-03) · icebreaker-ai/docs/v51/integrity.json @ origin/main aa461acc (2026-10-04) · icebreaker-ai/NHL-experiment-log.md @ origin/main a859ba2c (2026-10-04) · icebreaker-ai/research/v51/goalie_rules.py @ origin/main f6098309 (2026-10-03)
xG Underperformer Bounce is forward-only, not Table 2 (guide v3.5.0): misses proof, clv. It stays frozen and stamped (0 forward fires so far); it is listed as paper only if it passes the one-miss route with a positive CLV, on Kalshi or Polymarket venue CLV for >= 100 forward bets.
T_HC_NEW is forward-only, not Table 2 (guide v3.5.0): misses clv · CLV wrong-signed. It stays frozen and stamped (5 forward fires so far); it is listed as paper only if it passes the one-miss route with a positive CLV, on Kalshi or Polymarket venue CLV for >= 100 forward bets.
GK_STARTER_DOG is forward-only, not Table 2 (guide v3.5.0): frozen 2026-10-03T23:01:00Z (goalie_rules.py), no sealed set (forward bets do not supply criterion 1), no paper rows. Stamped so far: 0 forward fires. It is never counted as Table 1 or 2 until it clears the gate on its own forward bets.
GK_LATE_SWAP is forward-only, not Table 2 (guide v3.5.0): frozen 2026-10-03T23:01:00Z (goalie_rules.py), no sealed set (forward bets do not supply criterion 1), no paper rows. Stamped so far: 0 forward fires. It is never counted as Table 1 or 2 until it clears the gate on its own forward bets.
2 of 2 registered paper rules listed (counted in icebreaker-ai nhl_system_registry.json + research/v51/grade_forward_frozen.py (RULES)).
Table 1 live (1)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| Fade Overtime Hangover | 170 | 14.1% | 0.032 | 1.2% | 3/3 | 886 · 1.86% · <0.001 | 716/150 · full blind seasons 0/2 (≥ 75) · 0.6% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a single-archive blind window (NHL 2008-19, guide v3.8.0, 0.5% penalty); the increment over plain back-to-back opponents is not significant (random-subset p .09 sealed / .23 blind); Kalshi 2025-26 at ask+fee +3.76% on 94 — real-price results are tracked from now. Sealed 102-68-0 (+23.9u) · Blind 400-316-0 (+4.0u) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (2)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| Rested Small Fav vs B2B | 225 | 5.9% | 0.157 | -3.5% | 2/3 | 225 · 3.96% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 5 stamped rows, file 2026-10-03 17:43 UTC icebreaker-ai/data/forward/frozen_rule_fires_2026-27.jsonl (all four frozen rules write one file) |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 136-89-0 (+13.2u) · Blind 0-0-0 (+0.0u) forward: 0 stamped / 150 · frozen 2026-10-02 | |||||||||
| Rested Process Fav vs B2B | 107 | 5.5% | 0.257 | -8.0% | 2/3 | 107 · 4.34% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 5 stamped rows, file 2026-10-03 17:43 UTC icebreaker-ai/data/forward/frozen_rule_fires_2026-27.jsonl (all four frozen rules write one file) |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
same family as Rested Small Fav vs B2B: every one of its sealed picks is also that rule's, so the family is graded once and counted once Sealed 65-42-0 (+5.9u) · Blind 0-0-0 (+0.0u) forward: 0 stamped / 150 · frozen 2026-10-02 | |||||||||
NBA
Closest to a seat: **4 Table 1 seats live (live /v51 read 2026-10-04 20:35Z, 0 unverified): t02109 UNDER (2012-18 proxy blind, 631 bets +3.52%), s2838, s5200, o0090 (back the away team when the home team's last game within 3 days went to overtime; 2008-19 proxy blind 673 bets +1.57% at the worse price; criterion 2 sealed 2 of 3, owner 2026-10-04 "Keep rule as written"; thin: sealed edge mostly 2024-25, increment over the 3-day parent p .137)**. 4 live seats lower NBA's priority (owner 2026-10-03). Blowout-rest killed 2026-10-05 (courtside 8022784; 252 rules, 0 families; blind 2008-19 / 2012-18 unread). Injury-report vs open: not provable today (ESPN opens carry no clock; no priced blind window with reports) — forward-blind only. Next: /fresh-sports-idea-hunt nba before the 10-20 opener.; 7 paper rules (5 fresh + AGREE/DEFENSE + t11073, a right-signed CLV one-miss: combined mean +0.83%, p .0545), blind 0/150 on the real-price forward arm (season opens 2026-10-20)
Biggest obstacle: No real past opening price for spread or total (2012-18 open total price is -110 on 97.8-99.9%); 2012-18 moneyline window spent; the totals search is noise-compatible against an efficient-close null (P .555); the cross-book open gap (ESPN open vs Action Network b30 open, totals + spreads, 48 rules, 2023-26) is killed 2026-10-03: 0 passers, engine B exact, the gap only predicts the close in ESPN BET's stale 2024-25 open (courtside-ai `docs/preregistration/nba-v51-open-gap.md`); next: venue-vs-book mid gap (forward only)
Top leads
Forward stamps + Kalshi/Polymarket close capture (totals contracts now read too, exact strike only); **RLMDOG-NBA** (RLM underdog at T-2h, venue fire-time price) is forward-only, not Table 2: Polymarket fails CLV + seasons + blind, Kalshi fails proof + seasons + blind; fire stamps live from 2026-10-03; earliest Table 1 = end of the 2026-27 season (150 blind bets ~mid-Feb 2027 and the 3rd season complete)
Read from: courtside-ai/shared/nba-systems-contract.json @ origin/main 0f09890a (2026-09-30) · courtside-ai/scripts/analysis/nba_v51_agree_open.json @ origin/main eb455459 (2026-10-01) · courtside-ai/scripts/analysis/nba_v51_fresh_restated.json @ origin/main 7b724760 (2026-10-03) · courtside-ai/data/forward/nba_fresh_open_v1/rules.json @ origin/main f6de58fd (2026-10-02) · courtside-ai/data/forward/nba_totals_open_v1/rules.json @ origin/main 3581a59e (2026-10-03) · courtside-ai/scripts/analysis/nba_totals_window_candidates.json @ origin/main 38c54b8c (2026-10-03) · courtside-ai/scripts/analysis/nba_proxy_totals_blind.json @ origin/main 820ba39d (2026-10-04) · courtside-ai/data/forward/nba_totals_open_v1/seats.json @ origin/main cfba1831 (2026-10-04) · courtside-ai/data/forward/nba_minutes_spread_open_v1/rules.json @ origin/main 12dec268 (2026-10-04) · courtside-ai/data/forward/nba_minutes_spread_open_v1/seats.json @ origin/main 12dec268 (2026-10-04) · courtside-ai/scripts/analysis/nba_minutes_spread_blind.json @ origin/main 7d9a80f6 (2026-10-04) · courtside-ai/scripts/analysis/nba_minutes_spread_candidates.json @ origin/main a4d69438 (2026-10-04) · courtside-ai/data/forward/nba_ot_open_v1/rules.json @ origin/main 5c2874cf (2026-10-04) · courtside-ai/data/forward/nba_ot_open_v1/seats.json @ origin/main a45ecbc7 (2026-10-04) · courtside-ai/scripts/analysis/nba_ot_blind.json @ origin/main 994caf64 (2026-10-04) · courtside-ai/scripts/analysis/nba_ot_candidates.json @ origin/main 75d411c4 (2026-10-04)
Forward stamps (data/forward/nba_*) are written on the iMac and are not in git; blind counts are the registry's.
7 of 7 registered paper rules listed (counted in courtside-ai data/forward/nba_fresh_open_v1/rules.json + data/forward/nba_totals_open_v1/rules.json + data/forward/nba_minutes_spread_open_v1/rules.json + data/forward/nba_ot_open_v1/rules.json + scripts/analysis/nba_v51_agree_open.json).
Table 1 live (4)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| t02109 | 215 | 11.3% | 0.040 | 0.9% | 2/3 | 846 · 2.96% · <0.001 | 631/150 · full blind seasons 0/2 (≥ 75) · 3.5% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2012-18, guide v3.7.0); the proxy averages -106 vs -110 at real opens — at flat -110 the blind ROI is still +1.70%; zero-skill false-pass rate 28-47%, 3 of 6 blind seasons negative — real-price results are tracked from 10-20. Sealed 125-88-2 (+24.3u) · Blind 333-292-6 (+22.2u) forward: — stamped / 150 · frozen 2026-10-03 (first run pending 2026-10-20) | |||||||||
| o0090 | 116 | 17.4% | 0.024 | 3.4% | 2/3 | 789 · 0.87% · 0.008 | 673/150 · full blind seasons 0/2 (≥ 75) · 1.6% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2008-19, guide v3.7.0), graded at the worse of the proxy and a real open (-109.3): blind ROI +1.57% on 673 bets (at the proxy alone +3.13%), 6 of 11 blind seasons positive (2008-16 lost -4.2% on 506 bets, 2016-19 won). Blind seasons, bets / ROI: 08-09 62 / -10.8%, 09-10 66 / -8.6%, 10-11 66 / +2.5%, 11-12 54 / -2.5%, 12-13 64 / +9.1%, 13-14 60 / -9.0%, 14-15 59 / +3.8%, 15-16 75 / -15.9%, 16-17 64 / +17.8%, 17-18 55 / +23.4%, 18-19 48 / +15.8%. Criterion 2 passes on the 3 sealed seasons only (2 of 3: +9.7%, +56.7%, -1.8%); counting the 11 blind seasons too it would be 8 of 14 (fails). The sealed edge came mostly from one season (2024-25); its increment over the plain 3-day-rest parent is p .137; zero-skill replays clear this blind bar 6-11.5% of the time. Real-price results are tracked from 10-20. Sealed 70-43-3 (+20.1u) · Blind 350-308-15 (+10.6u) forward: — stamped / 150 · frozen 2026-10-04 (first run pending 2026-10-20) | |||||||||
| s2838 | 107 | 18.5% | 0.021 | 3.8% | 3/3 | 593 · 2.50% · <0.001 | 486/150 · full blind seasons 0/2 (≥ 75) · 1.5% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2008-19, guide v3.7.0), graded at the worse of the proxy and a real open (-109.3); blind ROI +1.51% on 486 bets (at the proxy alone +2.92%); zero-skill replays clear this blind bar 17-35% of the time, 7 of 11 blind seasons positive — real-price results are tracked from 10-20. Sealed 65-40-2 (+19.8u) · Blind 252-222-12 (+7.3u) forward: — stamped / 150 · frozen 2026-10-04 (first run pending 2026-10-20) | |||||||||
| s5200 | 161 | 13.6% | 0.034 | 1.6% | 2/3 | 596 · 2.44% · <0.001 | 435/150 · full blind seasons 0/2 (≥ 75) · 0.2% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Table 1 on a proxy-priced blind window (2008-19, guide v3.7.0), graded at the worse of the proxy and a real open (-109.3); blind ROI +0.16% on 435 bets (at the proxy alone +1.51%); zero-skill replays clear this blind bar 17-35% of the time, 4 of 11 blind seasons positive — real-price results are tracked from 10-20. Sealed 95-64-2 (+21.9u) · Blind 224-203-8 (+0.7u) forward: — stamped / 150 · frozen 2026-10-04 (first run pending 2026-10-20) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (7)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| nba-v51-agree-defense-open | 372 | 9.6% | 0.025 | 1.6% | 3/3 | 209 · 0.39% · 0.234 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_agree_defense_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✗ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 214-154-4 (+35.6u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · freeze date not recorded (first run pending 2026-10-20) | |||||||||
| r6330 | 333 | 10.4% | 0.023 | 1.9% | 2/3 | 333 · 4.60% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_fresh_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 193-140-0 (+34.7u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-20) | |||||||||
| r0227 | 310 | 11.1% | 0.019 | 2.3% | 2/3 | 310 · 6.49% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_fresh_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 180-129-1 (+34.5u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-20) | |||||||||
| r6240 | 267 | 10.1% | 0.042 | 0.5% | 2/3 | 267 · 5.27% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_fresh_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 154-113-0 (+26.8u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-20) | |||||||||
| r4425 | 264 | 12.0% | 0.020 | 2.5% | 3/3 | 264 · 4.73% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_fresh_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 155-108-1 (+31.7u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-20) | |||||||||
| r6107 | 259 | 11.9% | 0.022 | 2.2% | 2/3 | 259 · 4.76% · <0.001 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_fresh_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 151-108-0 (+30.9u) · Blind 0-0-0 (+0.0u) forward: — stamped / 150 · frozen 2026-10-02 (first run pending 2026-10-20) | |||||||||
| t11073 | 127 | 14.0% | 0.046 | 0.6% | 2/3 | 527 · 0.83% · 0.055 | 400/150 · full blind seasons 0/2 (≥ 75) · 4.6% | origins yes · engines yes | first run pending (2026-10-20) courtside-ai/data/forward/nba_totals_open_v1/stamps.jsonl — NBA stampers start with the 2026-27 season, opening 2026-10-20 (docs/v51-table2-candidates.md) |
✓ Proof ✓ Seasons ✗ CLV ✓ Blind ✓ ROI band ✓ Integrity
one blind look taken on the 2012-18 proxy window (guide v3.7.0): 400 bets, ROI +4.62%; missed clv; stays paper Sealed 77-50-0 (+17.8u) · Blind 213-183-4 (+18.5u) forward: — stamped / 150 · frozen 2026-10-03 (first run pending 2026-10-20) | |||||||||
MLB
Closest to a seat: **1 Table 1 seat: Back Home Win Streak (seated 2026-10-04, guide v3.11.0, method 1 SBRO 2010-15: blind 165 +5.83%, seasons 5/6 strict); live=False until the MLB ticket intent is built (first item of the next MLB Opus session; next slate 2027-03-25)**
Biggest obstacle: Frames repaired 2026-10-03 (384 Cleveland games had been missing from both engines 2018-20); 2018-20 seals pre-date the repair; no real opening price before 2018 except the one spent 2016 look
Top leads
5 other paper rules, all one-miss (4 on CLV, 1 on proof); Windy Even Matchup Dog blind +0.61% on 31
Read from: diamondpredictions/audit_reports/seal_corrections.jsonl @ origin/main 739be89e (2026-10-03) · diamondpredictions/audit_reports/v51_gate_report_restated_20261003.json @ origin/main fcd4c00e (2026-10-03) · diamondpredictions/audit_reports/v51_backward_2016_daily_rows_restated_20261003.json @ origin/main fcd4c00e (2026-10-03) · diamondpredictions/audit_reports/v51_gate_report.json @ origin/main 796f5dc4 (2026-10-04)
Out after the restated_20261003 correction (cle-alias-drop-restated_20261003): Cold Division Win Streak.
Fade Cold Revenge Mashers is forward-only, not Table 2 (guide v3.5.0): CLV miss with the wrong sign: mean CLV -0.43% is not > 0 (a fail, not a near-miss). It stays stamped in the MLB forward ledger (0 forward fires) and is listed as paper only if it later passes with a positive CLV.
Fade Windy Mashing Team is forward-only, not Table 2 (guide v3.5.0): CLV miss with the wrong sign: mean CLV -1.67% is not > 0 (a fail, not a near-miss). It stays stamped in the MLB forward ledger (0 forward fires) and is listed as paper only if it later passes with a positive CLV.
Back Home Win Streak sits in Table 1 on guide v3.11.0 (owner 2026-10-04). Its proof is the sealed 2018-25 record (137 bets, +15.3%, p 0.025) and book CLV +1.1% (p 0.00005, 302 bets). The blind record is 165 bets at +5.8%: 142 from MLB 2010-15, 23 from 2026. 2010-15 has one price archive (SBRO) and no second source, so the owner approved it with a penalty: a game whose opening price sits more than 15 points from its close is dropped, every payout is cut 0.5%, and live decisions always need two real prices. Read the blind step with care: a random rule passes it about half the time (95 of 200 trials at fair odds; a shuffle of results passes it 198 of 200), so the seat rests on the sealed proof and CLV, not on the blind ROI. Betting every home team at the 2010-15 open lost 3.5%, and random 144-bet home subsets matched this rule's 2010-15 result 10.5% of the time. 2016 (32 bets, +13.0%) is left out: the plan counted it, the seat does not, a cautious change made after the look (with it: 197 blind bets at +7.0%). Engines A and B agree on the sealed frame (137 bets) and on 2010-15 from 2010-05-01 (142 bets); the whole window did not (144 vs 147, three April-2010 bets where 2009 history is partial). Result note: diamondpredictions audit_reports/v51_single_archive_2010_15_result.md.
Paper rules are read from the latest correction's restated reports (restated_20261003); the un-restated originals stay on main unchanged.
3 of 3 registered paper rules listed (counted in diamondpredictions audit_reports/v51_gate_report_restated_20261003.json).
Table 1 live (1)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | |
|---|---|---|---|---|---|---|---|---|---|
| Back Home Win Streak | 137 | 15.3% | 0.025 | 2.2% | 5/6 | 302 · 1.11% · <0.001 | 165/150 · full blind seasons 0/2 (≥ 75) · 5.8% | origins yes · engines yes | |
✓ Proof ✓ Seasons ✓ CLV ✓ Blind ✓ ROI band ✓ Integrity Back Home Win Streak sits in Table 1 on guide v3.11.0 (owner 2026-10-04). Its proof is the sealed 2018-25 record (137 bets, +15.3%, p 0.025) and book CLV +1.1% (p 0.00005, 302 bets). The blind record is 165 bets at +5.8%: 142 from MLB 2010-15, 23 from 2026. 2010-15 has one price archive (SBRO) and no second source, so the owner approved it with a penalty: a game whose opening price sits more than 15 points from its close is dropped, every payout is cut 0.5%, and live decisions always need two real prices. Read the blind step with care: a random rule passes it about half the time (95 of 200 trials at fair odds; a shuffle of results passes it 198 of 200), so the seat rests on the sealed proof and CLV, not on the blind ROI. Betting every home team at the 2010-15 open lost 3.5%, and random 144-bet home subsets matched this rule's 2010-15 result 10.5% of the time. 2016 (32 bets, +13.0%) is left out: the plan counted it, the seat does not, a cautious change made after the look (with it: 197 blind bets at +7.0%). Engines A and B agree on the sealed frame (137 bets) and on 2010-15 from 2010-05-01 (142 bets); the whole window did not (144 vs 147, three April-2010 bets where 2009 history is partial). Result note: diamondpredictions audit_reports/v51_single_archive_2010_15_result.md. Sealed 86-51-0 (+20.9u) · Blind 98-67-0 (+9.6u) | |||||||||
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (3)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| Windy Even Matchup Dog | 138 | 20.7% | 0.015 | 4.8% | 4/5 | 169 · 0.35% · 0.283 | 31/150 · full blind seasons 0/2 (≥ 75) · 0.6% | passed (one flag) | 2 stamped rows, file 2026-10-05 05:19 UTC diamondpredictions/data/forward/mlb_v51_paper/stamps.jsonl |
✓ Proof ✓ Seasons ✗ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 74-64-0 (+28.5u) · Blind 14-17-0 (+0.2u) forward: 0 stamped / 150 · frozen 2026-10-03 | |||||||||
| Cold Division Slump Opponent | 108 | 20.7% | 0.010 | 6.6% | 6/6 | 162 · 0.30% · 0.274 | 54/150 · full blind seasons 0/2 (≥ 75) · 9.8% | passed (one flag) | 2 stamped rows, file 2026-10-05 05:19 UTC diamondpredictions/data/forward/mlb_v51_paper/stamps.jsonl |
✓ Proof ✓ Seasons ✗ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 72-36-0 (+22.4u) · Blind 33-21-0 (+5.3u) forward: 0 stamped / 150 · frozen 2026-10-03 | |||||||||
| Stingy Pitching Team | 200 | 10.0% | 0.090 | -2.5% | 4/6 | 288 · 0.94% · 0.023 | 88/150 · full blind seasons 0/2 (≥ 75) · 6.0% | passed (one flag) | 2 stamped rows, file 2026-10-05 05:19 UTC diamondpredictions/data/forward/mlb_v51_paper/stamps.jsonl |
✗ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Sealed 105-95-0 (+19.9u) · Blind 46-42-0 (+5.3u) forward: 0 stamped / 150 · frozen 2026-10-03 | |||||||||
Tennis
Closest to a seat: V51-OPEN-2 `OPEN2_PRICE:P3+GAP:G3+LEAD:L2`: 220 bets at +14.48% (p .0125, CI-lo +4.05, CLV +1.56% p .0192 n 220, seasons 2/3), frozen 9e0bae1; engines A and B agree (220 / +14.4773% both) and the 2021-23 opening price rests on the guide's pre-2024 single-archive exception (all 220 closes dual-verified, every open within 15pp of its close). Needs 150 blind bets, 0 so far (blind = matches after 2026-10-02T18:45:15Z)
Biggest obstacle: Blind pace: 220 sealed bets over 3 seasons is about 73 a year (derived), so 150 blind bets is about 2 seasons; seasons are exactly 2/3 (one more losing season fails it); the search result sits inside its null (48.5% of 200 null draws found 2 or more passers); 2024+ prices need a second open origin
Top leads
The second rule `OPEN2_PRICE:P3+GAP:G6+LEVEL:GS` (114 bets, +15.52%, p .0412) shares 86% of its picks with the first, so it is the same bet; Kalshi/Polymarket stamps and the blind grader are not written yet
Read from: tennis-algorithm/data/registry.json @ origin/main 6ef09eca (2026-10-01) · tennis-algorithm/data/backtest_results/v51_grade_2026-10-01.json @ origin/main 6ef09eca (2026-10-01) · tennis-algorithm/data/open_close/freeze_v51_open2.json @ origin/main 9e0bae10 (2026-10-02) · tennis-algorithm/data/open_close/blind_start_v51_open2.json @ origin/main b6f7bd8d (2026-10-02)
The V51-OPEN-2 forward capture is data/forward/tennis_open2/stamps.jsonl on the iMac; the 2026 venue files are unread.
2 of 2 registered paper rules listed (counted in tennis-algorithm data/open_close/freeze_v51_open2.json).
Table 1 live (0)
None live.
Table 1 approved, waiting to go live (0)
None waiting.
Table 2 paper (2)
| System | Sealed N | ROI at open | p | CI lower | Seasons | CLV (n · mean · p) | Blind (of 150, or 2 full seasons ≥ 75) · ROI | Integrity | Forward stamps (iMac) |
|---|---|---|---|---|---|---|---|---|---|
| OPEN2_PRICE:P3+GAP:G3+LEAD:L2 | 220 | 14.5% | 0.012 | 4.1% | 2/3 | 220 · 1.56% · 0.019 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 187 stamped rows, file 2026-10-05 15:03 UTC tennis-algorithm/data/forward/tennis_open2/stamps.jsonl (both V51-OPEN-2 rules write one file; stamped counts the rows where this rule's decision is fire) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 132-88-0 (+31.9u) · Blind 0-0-0 (+0.0u) forward: 3 stamped / 150 · frozen 2026-10-02 | |||||||||
| OPEN2_PRICE:P3+GAP:G6+LEVEL:GS | 114 | 15.5% | 0.041 | 1.1% | 2/3 | 114 · 1.94% · 0.042 | 0/150 · full blind seasons 0/2 (≥ 75) · — | origins yes · engines yes | 187 stamped rows, file 2026-10-05 15:03 UTC tennis-algorithm/data/forward/tennis_open2/stamps.jsonl (both V51-OPEN-2 rules write one file; stamped counts the rows where this rule's decision is fire) |
✓ Proof ✓ Seasons ✓ CLV … Blind ✓ ROI band ✓ Integrity
Derived from the values with the v5.1 gate thresholds (the artifact carries no verdict): Proof, Seasons, CLV, Blind, ROI band. Every other mark is the sport's own verdict. Sealed 69-45-0 (+17.7u) · Blind 0-0-0 (+0.0u) forward: 0 stamped / 150 · frozen 2026-10-02 | |||||||||