Ledger

Every game the model scored, graded against the final score and the market: one unit on the model's lean each game. Judge each ROI against its null (about −4%, the market's hold), not zero, and against the market favourite on the same games. Forward rows and rebuilt history are never pooled.

How to read these numbers

Judge every ROI against its null: the ROI if the market's no-vig prices were exactly right. That is about −4% (the hold: Kalshi's taker fee and spread, or the bookmaker's margin), not zero. Market is Kalshi's last pre-tip price wherever a game has one (2025-26 on), else the sportsbook close; the hypotheses stay on the sportsbook prices they were registered on. ± is one standard error (1 SE). Colour marks only gaps of at least 2 SE from the null: green above, red below; everything else is left plain. Shaded columns are the ones to read first.

Glossary

Lean: The side the model gives ≥ 50% (as recorded on the row).
Value side: The side where the model's probability beats the market's no-vig probability. Graded only through the pre-registered hypotheses; the blanket value bet is retired (see the note under the hypotheses).
Model %: The picked side's mean model win probability.
Market %: The picked side's mean no-vig market probability (q): the price with the bookmaker's margin removed.
Win %: How often the picked side actually won.
Break-even: The win rate the posted price needs to pay back; above q by about the hold.
ROI: Profit per 1-unit bet.
Null: The ROI expected if the market is right (q × payout − 1). vs null = ROI − null.
z: (ROI − null) ÷ SE. Across many cells, about one in twenty shows |z| ≥ 2 by chance.
EV: Win % − break-even; its null is q − break-even.
Fav ROI: 1u on the market favourite of the same games at the same price: the same-row baseline a pick rule has to beat (a tie at 50% goes to home).
Log loss, Brier: Proper scores of the probabilities; lower is better. Compared with the close on the same games.
Native: Written before tip and frozen: the forward test.
Reconstructed: Scored afterwards with that season left out of the fit and graded at the close: hindsight, never forward evidence.

Pre-registered hypotheses — the forward test

Rules frozen before any native rows; every result is reported here, win or lose. Native is the forward test; hindsight is the same rule on the rebuilt seasons.

	Rule	Native n	ROI	Null	vs null (± 1 SE)	Hindsight n	ROI	Null	vs null (± 1 SE)
H1	Games 1–9 value side · model P − no-vig q ≥ 0.08	0 at the pregame	—	—	—	115 at the close	+17.8%	-4.0%	+21.8 ± 12.8
H2	Games 10+ value side that is the favourite	0 at the pregame	—	—	—	987 at the open	+0.6%	-4.1%	+4.7 ± 2.2
H3	Games 10+ value side · model P − no-vig q ≥ 0.12	0 at the pregame	—	—	—	260 at the open	+19.0%	-4.2%	+23.2 ± 8.7
H4	Games 10+ value side · model P − no-vig q ≥ 0.12	0 at the first snapshot	—	—	—	260 at the open	+19.0%	-4.2%	+23.2 ± 8.7
H2·F	Games 10+ value side that is the favourite	0 at the first snapshot	—	—	—	987 at the open	+0.6%	-4.1%	+4.7 ± 2.2

About these hypotheses — dates, prices, what was retired

H1–H3 fixed on 2026-09-30, H4 and amendment A1 (H2·F: H2's rule at the first snapshot) on 2026-10-01, all before any native rows; the thresholds are frozen and every result is reported here, win or lose. Native bets are graded at a price that could have been bet: the pregame snapshot is the last one before tip (near the close); the first snapshot is the earliest priced one, with the model probability written then (H4 is H3's rule at the early price, where the hindsight edge appeared). Hindsight is the same rule on the reconstructed rows at the price the scan found it (pooled over both seasons and books, as registered), recomputed on the current reconstructed model; one season of native rows is not a verdict.

Retired 2026-10-06: the blanket value bet (every game, the side with model P > no-vig q) and its ATS twin. On the reconstructed rows at the close its claimed EV was +11 to +16% in every book and season while it returned −6 to −11% (null about −4%; ATS −3 to −7% vs −4.5%). The value-side rules still under test are the frozen hypotheses in this table.

Forward record Native · forward

No native rows yet. The daily build writes one before each tip, starting with the first regular-season slate; this section fills in as those games are graded.

Rebuilt history Reconstructed · hindsight

The current model replayed on past seasons, each season left out of its own fit and graded at the close. Hindsight, not a track record: the model was designed after these games were played.

ESPN BET close · 2024-25825–384 · ROI -3.5% · null -4.2% · fav -4.3%

Scored afterwards with this season left out of the fit · 1209 graded games on 159 slates.

Probabilities vs the close (log loss)

Market better

by 2.4 SE · model − market +0.0118 ± 0.0050 (1 SE)
0.5940 vs 0.5821 · n=1209

Lean bets · 1u at the close

-3.5% ROI

vs null (-4.2%): +0.7 ± 2.0 — not resolved (z = +0.3)
825–384 · won 68.2%, break-even 70.8%
market fav, same games: -4.3%

ROI — one unit on every pick

At the closing price.

Lean — the model's favourite

Picks	n	W–L	ROI	Null	vs null (± 1 SE)	Win %	Break-even	Model %	Market %	Fav ROI
All picks	1209	825–384	-3.5%	-4.2%	+0.7 ± 2.0	68.2%	70.8%	69.1%	67.8%	-4.3%
Games 10+	1071	732–339	-3.9%	-4.2%	+0.3 ± 2.1	68.3%	71.0%	69.0%	68.0%	-3.6%
Games 1–9 (carryover)	138	93–45	-0.4%	-4.2%	+3.8 ± 6.5	67.4%	69.5%	69.4%	66.6%	-10.1%

ROI by price band — one unit on every pick, by the picked side's moneyline

Graded at the closing price. With 8 cells per section, a |z| near 2 somewhere happens by chance about one section in three. Bands are descriptive monitoring dimensions, not a filter: a band that looks good here is a hypothesis to test forward, and the hypotheses under test are the pre-registered ones at the top of this page (the earlier DraftKings −249 to −130 lean band is retired).

Lean — the model's favourite

Picked price	n	W–L	ROI	Null	vs null (± 1 SE)	z	Win %	Break-even	EV (pp)	EV null (pp)	Model %	Market %
≤ -250	594	467–127	-5.0%	-4.4%	-0.5 ± 2.1	-0.3	78.6%	82.5%	-3.9	-3.7	77.4%	78.9%
-249 to -175	243	156–87	-4.5%	-4.0%	-0.5 ± 4.6	-0.1	64.2%	67.2%	-3.0	-2.7	64.8%	64.6%
-174 to -130	195	118–77	+1.1%	-3.9%	+5.0 ± 5.9	+0.9	60.5%	59.9%	+0.6	-2.3	60.4%	57.6%
-129 to -100	87	47–40	+0.8%	-4.4%	+5.2 ± 10.0	+0.5	54.0%	53.6%	+0.5	-2.3	57.5%	51.2%
+100 to +129	54	22–32	-13.0%	-4.1%	-8.9 ± 14.4	-0.6	40.7%	47.3%	-6.5	-1.9	55.9%	45.3%
+130 to +174	24	13–11	+33.8%	-3.9%	+37.6 ± 25.7	+1.5	54.2%	40.5%	+13.7	-1.6	55.3%	38.9%
+175 to +249	10	2–8	-43.5%	-4.0%	-39.5 ± 37.7	-1.0	20.0%	34.1%	-14.1	-1.4	56.0%	32.7%
≥ +250	2	0–2	-100.0%	-3.8%	-96.2 ± 0.0	—	0.0%	28.2%	-28.2	-1.1	59.6%	27.1%

Against the spread — one unit on every pick at the closing spread

Lean ATS — the model's moneyline favourite against the spread

Picks	n	W–L–P	ROI	Null	vs null (± 1 SE)	Cover %	Break-even	Model cover %	Market cover %
All picks	1208	606–602–0	-4.3%	-4.5%	+0.2 ± 2.7	50.2%	52.4%	52.2%	50.0%

Only rows with a recorded closing spread (same book as the moneyline close). Model cover % maps the model's win probability to a margin with σ = 13.5 pts; it is a display mapping, not a new prediction. Cover % is covers ÷ (covers + misses); pushes refund. The null is minus the spread hold.

Game by game — latest 60 of 1209, newest first

Date	Away @ Home	Lean	Model %	Market %	Close ML	Final (away–home)	Lean · P/L (1u)
2025-04-13	DEN @ HOU	DEN	76.6%	67.6%	-240	126–111	W +0.42u
2025-04-13	DAL @ MEM	MEM	59.2%	44.7%	+115	97–132	W +1.15u
2025-04-13	UTA @ MIN	MIN	96.6%	95.4%	-7500	105–116	W +0.01u
2025-04-13	OKC @ NOP	OKC	82.2%	84.2%	-800	115–100	W +0.12u
2025-04-13	TOR @ SAS	SAS	60.5%	61.4%	-175	118–125	W +0.57u
2025-04-13	LAC @ GSW	GSW	56.9%	60.2%	-170	124–119	L -1.00u
2025-04-13	LAL @ POR	LAL	56.1%	34.9%	+175	81–109	L -1.00u
2025-04-13	PHO @ SAC	SAC	73.1%	86.4%	-1000	98–109	W +0.10u
2025-04-13	ORL @ ATL	ATL	68.0%	39.8%	+140	105–117	W +1.40u
2025-04-13	CHO @ BOS	BOS	95.2%	81.1%	-600	86–93	W +0.17u
2025-04-13	NYK @ BRK	NYK	73.3%	63.0%	-190	113–105	W +0.53u
2025-04-13	IND @ CLE	CLE	74.2%	32.1%	+200	126–118	L -1.00u
2025-04-13	WAS @ MIA	MIA	90.4%	74.8%	-360	119–118	L -1.00u
2025-04-13	CHI @ PHI	CHI	65.2%	73.3%	-320	122–102	W +0.31u
2025-04-13	DET @ MIL	DET	52.1%	76.9%	-400	133–140	L -1.00u
2025-04-11	HOU @ LAL	LAL	81.8%	88.1%	-1200	109–140	W +0.08u
2025-04-11	SAS @ PHO	PHO	68.2%	67.3%	-230	98–117	W +0.43u
2025-04-11	GSW @ POR	GSW	82.8%	88.8%	-1400	103–86	W +0.07u
2025-04-11	LAC @ SAC	LAC	64.8%	70.4%	-275	101–100	W +0.36u
2025-04-11	OKC @ UTA	OKC	92.4%	76.2%	-400	145–111	W +0.25u
2025-04-11	BRK @ MIN	MIN	93.1%	93.2%	-4000	91–117	W +0.03u
2025-04-11	MEM @ DEN	DEN	76.7%	71.7%	-290	109–117	W +0.34u
2025-04-11	TOR @ DAL	DAL	60.2%	81.1%	-600	102–124	W +0.17u
2025-04-11	WAS @ CHI	CHI	81.4%	85.4%	-900	89–119	W +0.11u
2025-04-11	MIA @ NOP	MIA	84.6%	86.4%	-1000	153–104	W +0.10u
2025-04-11	CHO @ BOS	BOS	90.5%	93.2%	-4000	94–130	W +0.03u
2025-04-11	CLE @ NYK	NYK	58.4%	58.8%	-155	108–102	L -1.00u
2025-04-11	MIL @ DET	MIL	50.3%	40.1%	+140	125–119	W +1.40u
2025-04-11	ORL @ IND	IND	70.7%	60.7%	-170	129–115	L -1.00u
2025-04-11	ATL @ PHI	ATL	67.9%	83.5%	-750	124–110	W +0.13u
2025-04-10	MIN @ MEM	MIN	61.1%	54.3%	-130	141–125	W +0.77u
2025-04-10	NOP @ MIL	MIL	92.3%	89.4%	-1600	111–136	W +0.06u
2025-04-10	ATL @ BRK	ATL	73.4%	84.2%	-800	133–109	W +0.12u
2025-04-10	NYK @ DET	NYK	51.5%	40.9%	+135	106–115	L -1.00u
2025-04-10	CLE @ IND	IND	64.1%	78.3%	-450	112–114	W +0.22u
2025-04-09	HOU @ LAC	LAC	54.3%	81.1%	-600	117–134	W +0.17u
2025-04-09	SAS @ GSW	GSW	83.9%	90.5%	-1800	114–111	L -1.00u
2025-04-09	OKC @ PHO	OKC	78.7%	70.0%	-270	125–112	W +0.37u
2025-04-09	DEN @ SAC	DEN	58.1%	63.0%	-190	124–116	W +0.53u
2025-04-09	POR @ UTA	POR	71.2%	68.9%	-265	126–133	L -1.00u
2025-04-09	MIA @ CHI	MIA	59.5%	53.7%	-130	111–119	L -1.00u
2025-04-09	CHO @ TOR	TOR	77.1%	80.2%	-550	96–126	W +0.18u
2025-04-09	LAL @ DAL	LAL	67.3%	63.9%	-200	112–97	W +0.50u
2025-04-09	BOS @ ORL	BOS	63.7%	26.6%	+260	76–96	L -1.00u
2025-04-09	PHI @ WAS	PHI	62.8%	57.1%	-145	122–103	W +0.69u
2025-04-08	SAS @ LAC	LAC	84.0%	83.5%	-750	117–122	W +0.13u
2025-04-08	GSW @ PHO	GSW	67.8%	78.6%	-450	133–95	W +0.22u
2025-04-08	MIN @ MIL	MIN	70.3%	65.8%	-220	103–110	L -1.00u
2025-04-08	LAL @ OKC	OKC	79.4%	76.9%	-400	120–136	W +0.25u
2025-04-08	NOP @ BRK	BRK	53.9%	51.1%	-115	114–119	W +0.87u
2025-04-08	BOS @ NYK	BOS	61.2%	55.3%	-135	119–117	W +0.74u
2025-04-08	MEM @ CHO	MEM	80.9%	88.1%	-1200	124–100	W +0.08u
2025-04-08	CHI @ CLE	CLE	85.3%	88.1%	-1200	113–135	W +0.08u
2025-04-08	WAS @ IND	IND	93.2%	92.6%	-3000	98–104	W +0.03u
2025-04-08	ATL @ ORL	ORL	55.8%	64.3%	-200	112–119	W +0.50u
2025-04-07	PHI @ MIA	MIA	77.7%	80.8%	-525	105–117	W +0.19u
2025-04-07	SAC @ DET	DET	58.8%	63.0%	-190	127–117	L -1.00u
2025-04-06	HOU @ GSW	GSW	54.7%	65.8%	-220	106–96	L -1.00u
2025-04-06	MIL @ NOP	MIL	71.9%	66.9%	-230	111–107	W +0.43u
2025-04-06	IND @ DEN	DEN	59.9%	67.9%	-240	125–120	L -1.00u

The full record is in the CSV under data/.

Kalshi close · 2025-26834–375 · ROI -0.9% · null -3.2% · fav -3.3%

Scored afterwards with this season left out of the fit · 1209 graded games on 160 slates.

Probabilities vs the close (log loss)

Market better

by 2.9 SE · model − market +0.0159 ± 0.0055 (1 SE)
0.5837 vs 0.5677 · n=1209

Lean bets · 1u at the close

-0.9% ROI

vs null (-3.2%): +2.3 ± 2.1 — not resolved (z = +1.1)
834–375 · won 69.0%, break-even 69.9%
market fav, same games: -3.3%

ROI — one unit on every pick

At the closing price.

Lean — the model's favourite

Picks	n	W–L	ROI	Null	vs null (± 1 SE)	Win %	Break-even	Model %	Market %	Fav ROI
All picks	1209	834–375	-0.9%	-3.2%	+2.3 ± 2.1	69.0%	69.9%	68.0%	67.8%	-3.3%
Games 10+	1071	739–332	-1.9%	-3.1%	+1.3 ± 2.2	69.0%	70.4%	68.3%	68.4%	-3.7%
Games 1–9 (carryover)	138	95–43	+6.7%	-3.5%	+10.2 ± 6.9	68.8%	65.4%	65.8%	63.3%	+0.0%

ROI by price band — one unit on every pick, by the picked side's moneyline

Lean — the model's favourite

Picked price	n	W–L	ROI	Null	vs null (± 1 SE)	z	Win %	Break-even	EV (pp)	EV null (pp)	Model %	Market %
≤ -250	594	480–114	-1.6%	-2.2%	+0.6 ± 2.0	+0.3	80.8%	82.0%	-1.2	-1.8	76.4%	80.3%
-249 to -175	218	143–75	-2.5%	-3.4%	+0.9 ± 4.8	+0.2	65.6%	67.1%	-1.5	-2.3	64.1%	64.8%
-174 to -130	185	114–71	+2.9%	-4.0%	+6.9 ± 6.0	+1.2	61.6%	59.8%	+1.8	-2.4	59.1%	57.4%
-129 to -100	87	38–49	-18.2%	-4.5%	-13.7 ± 10.0	-1.4	43.7%	53.5%	-9.9	-2.4	56.2%	51.2%
+100 to +129	74	38–36	+8.8%	-5.0%	+13.7 ± 12.4	+1.1	51.4%	47.2%	+4.2	-2.4	56.7%	44.8%
+130 to +174	36	14–22	-4.5%	-5.6%	+1.1 ± 20.3	+0.1	38.9%	40.4%	-1.5	-2.3	55.1%	38.1%
+175 to +249	12	6–6	+53.8%	-6.5%	+60.2 ± 46.6	+1.3	50.0%	32.8%	+17.2	-2.1	54.2%	30.6%
≥ +250	3	1–2	+109.7%	-8.7%	+118.3 ± 209.7	+0.6	33.3%	18.7%	+14.7	-1.5	60.2%	17.1%

Against the spread — one unit on every pick at the closing spread

Lean ATS — the model's moneyline favourite against the spread

Picks	n	W–L–P	ROI	Null	vs null (± 1 SE)	Cover %	Break-even	Model cover %	Market cover %
All picks	1209	610–599–0	-3.6%	-4.5%	+0.9 ± 2.8	50.5%	52.3%	50.7%	50.0%

Game by game — latest 60 of 1209, newest first

Date	Away @ Home	Lean	Model %	Market %	Close ML	Final (away–home)	Lean · P/L (1u)
2026-04-12	CHI @ DAL	DAL	53.0%	64.5%	-199	128–149	W +0.50u
2026-04-12	MEM @ HOU	HOU	90.3%	89.5%	-967	101–132	W +0.10u
2026-04-12	NOP @ MIN	MIN	69.3%	60.5%	-168	126–132	W +0.60u
2026-04-12	PHO @ OKC	OKC	68.1%	72.8%	-306	135–103	L -1.00u
2026-04-12	DEN @ SAS	SAS	80.7%	79.3%	-404	128–118	L -1.00u
2026-04-12	UTA @ LAL	LAL	79.8%	90.5%	-1087	107–131	W +0.09u
2026-04-12	GSW @ LAC	LAC	61.3%	59.5%	-161	110–115	W +0.62u
2026-04-12	SAC @ POR	POR	80.4%	94.5%	-2042	110–122	W +0.05u
2026-04-12	ORL @ BOS	BOS	69.8%	14.5%	+529	108–113	W +5.29u
2026-04-12	WAS @ CLE	CLE	85.7%	80.5%	-458	117–130	W +0.22u
2026-04-12	DET @ IND	DET	84.5%	90.5%	-1087	133–121	W +0.09u
2026-04-12	ATL @ MIA	MIA	65.4%	78.7%	-430	117–143	W +0.23u
2026-04-12	CHO @ NYK	NYK	53.7%	12.4%	+625	110–96	L -1.00u
2026-04-12	MIL @ PHI	PHI	77.0%	92.4%	-1236	106–126	W +0.08u
2026-04-12	BRK @ TOR	TOR	90.2%	97.5%	-5268	101–136	W +0.02u
2026-04-10	PHO @ LAL	LAL	54.0%	44.5%	+114	73–101	W +1.14u
2026-04-10	LAC @ POR	POR	51.7%	53.5%	-126	97–116	W +0.79u
2026-04-10	GSW @ SAC	GSW	59.3%	81.5%	-489	118–124	L -1.00u
2026-04-10	MIN @ HOU	HOU	65.6%	80.5%	-458	136–132	L -1.00u
2026-04-10	MEM @ UTA	UTA	55.8%	63.5%	-191	101–147	W +0.52u
2026-04-10	OKC @ DEN	OKC	86.0%	43.9%	+119	107–127	L -1.00u
2026-04-10	ORL @ CHI	ORL	78.5%	90.5%	-1087	127–103	W +0.09u
2026-04-10	BRK @ MIL	MIL	73.7%	80.5%	-458	108–125	W +0.22u
2026-04-10	DAL @ SAS	SAS	90.9%	93.5%	-1684	120–139	W +0.06u
2026-04-10	NOP @ BOS	BOS	88.7%	94.4%	-1684	118–144	W +0.06u
2026-04-10	PHI @ IND	PHI	60.7%	88.6%	-967	105–94	W +0.10u
2026-04-10	TOR @ NYK	NYK	72.4%	70.5%	-263	95–112	W +0.38u
2026-04-10	DET @ CHO	DET	58.6%	35.1%	+166	118–100	W +1.66u
2026-04-10	MIA @ WAS	MIA	79.1%	91.5%	-1236	140–117	W +0.08u
2026-04-10	CLE @ ATL	CLE	57.2%	24.5%	+280	102–124	L -1.00u
2026-04-09	LAL @ GSW	GSW	59.8%	52.5%	-121	119–103	L -1.00u
2026-04-09	PHI @ HOU	HOU	76.2%	69.5%	-251	102–113	W +0.40u
2026-04-09	IND @ BRK	IND	57.6%	59.9%	-168	123–94	W +0.60u
2026-04-09	BOS @ NYK	NYK	53.8%	60.5%	-168	106–112	W +0.60u
2026-04-09	MIA @ TOR	TOR	59.0%	60.9%	-175	114–128	W +0.57u
2026-04-09	CHI @ WAS	CHI	54.5%	71.5%	-276	119–108	W +0.36u
2026-04-08	OKC @ LAC	OKC	77.5%	75.7%	-360	128–110	W +0.28u
2026-04-08	DAL @ PHO	PHO	80.9%	85.5%	-660	107–112	W +0.15u
2026-04-08	POR @ SAS	SAS	71.9%	58.9%	-161	101–112	W +0.62u
2026-04-08	MEM @ DEN	DEN	92.4%	97.5%	-5268	119–136	W +0.02u
2026-04-08	ATL @ CLE	CLE	62.2%	61.5%	-175	116–122	W +0.57u
2026-04-08	MIN @ ORL	ORL	61.5%	81.5%	-489	120–132	W +0.20u
2026-04-08	MIL @ DET	DET	91.1%	95.5%	-2580	111–137	W +0.04u
2026-04-07	HOU @ PHO	HOU	55.0%	52.5%	-121	119–105	W +0.83u
2026-04-07	OKC @ LAL	OKC	89.0%	94.5%	-2042	123–87	W +0.05u
2026-04-07	DAL @ LAC	LAC	85.4%	84.5%	-609	103–116	W +0.16u
2026-04-07	SAC @ GSW	GSW	82.2%	89.5%	-967	105–110	W +0.10u
2026-04-07	CHO @ BOS	BOS	72.6%	63.5%	-191	102–113	W +0.52u
2026-04-07	UTA @ NOP	NOP	69.2%	69.5%	-251	137–156	W +0.40u
2026-04-07	MIL @ BRK	MIL	57.9%	53.0%	-121	90–96	L -1.00u
2026-04-07	MIA @ TOR	TOR	57.0%	56.5%	-142	95–121	W +0.70u
2026-04-07	CHI @ WAS	CHI	53.5%	68.5%	-239	129–98	W +0.42u
2026-04-07	MIN @ IND	MIN	72.5%	86.5%	-719	124–104	W +0.14u
2026-04-06	POR @ DEN	DEN	76.2%	72.5%	-290	132–137	W +0.34u
2026-04-06	CLE @ MEM	CLE	74.0%	88.5%	-869	142–126	W +0.12u
2026-04-06	PHI @ SAS	SAS	81.9%	76.5%	-360	102–115	W +0.28u
2026-04-06	NYK @ ATL	NYK	55.0%	48.5%	-103	108–105	W +0.97u
2026-04-06	DET @ ORL	DET	63.6%	51.0%	-116	107–123	L -1.00u
2026-04-05	HOU @ GSW	HOU	70.5%	61.5%	-175	117–116	W +0.57u
2026-04-05	LAC @ SAC	LAC	81.0%	86.5%	-719	138–109	W +0.14u

The full record is in the CSV under data/.

Built 2026-10-19 07:42 UTC. Win probabilities from prior-game four factors, roster talent and the NBA injury report; the market is a benchmark only and never enters the model. Full statistics: ledger_report.txt · model tags and coefficients: Model. Not betting advice.