NFL Game Projections | Rusty Bill Sports

Research model, not a validated betting model. Scores, spreads, totals and win probabilities come from game-projection-v1.2 using football data only (nflverse play-by-play, injuries, starters, weather). Sportsbook lines are compared afterward and are never inputs. The official projection for each game is the latest one made before kickoff; it is locked at kickoff and graded. Prospective tracking started 2026 Week 7.

PROSPECTIVE / FROZEN-MODEL RESULTS

Official game-projection-v1.2 projections only: the latest projection made before each game's kickoff, locked at kickoff, graded after the final. These are the only results that count as out-of-sample.

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
Official projections	N = 29SMALL SAMPLE	0.4322	0.6249	65.5%	10.81	10.33	7.88	10.55 (model 10.81 on same 29)	10.48

Timing: early vs midweek vs late vs official

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
EARLY	N = 29SMALL SAMPLE	0.4353	0.6300	65.5%	11.04	10.32	8.03	10.55 (model 11.04 on same 29)	10.48
MIDWEEK	N = 29SMALL SAMPLE	0.4299	0.6225	62.1%	10.81	10.20	7.87	10.55 (model 10.81 on same 29)	10.48
LATE	N = 29SMALL SAMPLE	0.4422	0.6362	65.5%	11.10	10.20	7.96	10.55 (model 11.10 on same 29)	10.48
OFFICIAL	N = 29SMALL SAMPLE	0.4322	0.6249	65.5%	10.81	10.33	7.88	10.55 (model 10.81 on same 29)	10.48

Win-probability calibration (favorite's probability)

Favorite's final-game win probability	Games	Average model probability	Favorite actually won (tie = no)
45–60%	N = 14VERY SMALL SAMPLE	55.5%	57.1%
60–70%	N = 7VERY SMALL SAMPLE	64.4%	71.4%
70–80%	N = 7VERY SMALL SAMPLE	73.1%	71.4%
80–90%	N = 1VERY SMALL SAMPLE	81.8%	100.0%
90–100%	N = 0VERY SMALL SAMPLE	—	—

Extreme divergence from the closing line

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
Model vs closing spread differs by 4+ points	N = 6VERY SMALL SAMPLE	0.5321	0.7287	33.3%	14.29	9.43	8.34	11.08 (model 14.29 on same 6)	8.75
Win probability differs from closing no-vig by 8+ points	N = 13VERY SMALL SAMPLE	0.4897	0.6864	53.8%	9.04	12.43	8.48	8.27 (model 9.04 on same 13)	12.62

Inexperienced starting QB (under 300 career dropbacks)

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
Games with an inexperienced QB	N = 4VERY SMALL SAMPLE	0.2639	0.4297	75.0%	15.43	8.05	8.41	17.38 (model 15.43 on same 4)	8.38
All other games	N = 25SMALL SAMPLE	0.4592	0.6561	64.0%	10.07	10.70	7.79	9.46 (model 10.07 on same 25)	10.82

By model version

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
game-projection-v1.2	N = 29SMALL SAMPLE	0.4322	0.6249	65.5%	10.81	10.33	7.88	10.55 (model 10.81 on same 29)	10.48

Hypothetical Model Tracking (not actual bets)

1 unit on the model's preferred side at the best price available when the official projection was made. Research tracking only, not betting results and not recommendations.

Moneyline · line at projection time

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 29SMALL SAMPLE	12-17-0	41.4%	-1.8	-6.3%	+7.6	0.00 (N 29)
0-2% probability difference	N = 2VERY SMALL SAMPLE	0-2-0	0.0%	-2.0	-100.0%	+1.1	-0.00 (N 2)
2-4% probability difference	N = 6VERY SMALL SAMPLE	4-2-0	66.7%	+2.5	41.6%	+3.1	0.00 (N 6)
4-6% probability difference	N = 5VERY SMALL SAMPLE	3-2-0	60.0%	+0.4	8.1%	+4.8	0.01 (N 5)
6-8% probability difference	N = 3VERY SMALL SAMPLE	0-3-0	0.0%	-3.0	-100.0%	+6.9	-0.01 (N 3)
8%+ probability difference	N = 13VERY SMALL SAMPLE	5-8-0	38.5%	+0.3	2.0%	+11.9	0.00 (N 13)

Spread · line at projection time

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 29SMALL SAMPLE	16-13-0	55.2%	+2.5	8.7%	+8.1	0.07 (N 29)
0-2 pts difference	N = 17VERY SMALL SAMPLE	9-8-0	52.9%	+0.9	5.0%	+4.6	0.00 (N 17)
2-4 pts difference	N = 5VERY SMALL SAMPLE	4-1-0	80.0%	+2.8	55.2%	+11.9	0.40 (N 5)
4-6 pts difference	N = 6VERY SMALL SAMPLE	3-3-0	50.0%	-0.1	-1.6%	+13.0	0.00 (N 6)
6+ pts difference	N = 1VERY SMALL SAMPLE	0-1-0	0.0%	-1.0	-100.0%	+19.9	0.00 (N 1)
0-2% probability difference	N = 2VERY SMALL SAMPLE	0-2-0	0.0%	-2.0	-100.0%	+1.2	-0.25 (N 2)
2-4% probability difference	N = 4VERY SMALL SAMPLE	4-0-0	100.0%	+3.9	98.6%	+3.4	0.00 (N 4)
4-6% probability difference	N = 7VERY SMALL SAMPLE	3-4-0	42.9%	-0.9	-13.4%	+4.8	0.07 (N 7)
6-8% probability difference	N = 4VERY SMALL SAMPLE	2-2-0	50.0%	-0.1	-3.7%	+7.1	0.00 (N 4)
8%+ probability difference	N = 12VERY SMALL SAMPLE	7-5-0	58.3%	+1.7	13.8%	+13.1	0.17 (N 12)

Away team total · line at projection time

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 29SMALL SAMPLE	17-12-0	58.6%	+3.5	12.1%	+6.9	0.03 (N 29)
0-2 pts difference	N = 19VERY SMALL SAMPLE	11-8-0	57.9%	+1.9	10.1%	+4.6	0.11 (N 19)
2-4 pts difference	N = 10VERY SMALL SAMPLE	6-4-0	60.0%	+1.6	15.7%	+11.1	-0.10 (N 10)
0-2% probability difference	N = 4VERY SMALL SAMPLE	1-3-0	25.0%	-2.1	-52.3%	+0.9	0.00 (N 4)
2-4% probability difference	N = 3VERY SMALL SAMPLE	3-0-0	100.0%	+2.7	91.1%	+3.5	0.00 (N 3)
4-6% probability difference	N = 6VERY SMALL SAMPLE	4-2-0	66.7%	+1.6	26.2%	+4.9	0.00 (N 6)
6-8% probability difference	N = 7VERY SMALL SAMPLE	4-3-0	57.1%	+0.6	8.1%	+6.8	0.29 (N 7)
8%+ probability difference	N = 9VERY SMALL SAMPLE	5-4-0	55.6%	+0.7	7.9%	+12.0	-0.11 (N 9)

Home team total · line at projection time

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 29SMALL SAMPLE	17-12-0	58.6%	+4.4	15.1%	+6.1	0.03 (N 29)
0-2 pts difference	N = 20SMALL SAMPLE	13-7-0	65.0%	+5.9	29.4%	+4.3	-0.10 (N 20)
2-4 pts difference	N = 9VERY SMALL SAMPLE	4-5-0	44.4%	-1.5	-16.9%	+10.2	0.33 (N 9)
0-2% probability difference	N = 7VERY SMALL SAMPLE	4-3-0	57.1%	+1.1	15.6%	+0.9	0.00 (N 7)
2-4% probability difference	N = 2VERY SMALL SAMPLE	1-1-0	50.0%	-0.0	-1.0%	+3.6	-0.50 (N 2)
4-6% probability difference	N = 7VERY SMALL SAMPLE	4-3-0	57.1%	+0.9	13.3%	+5.1	-0.14 (N 7)
6-8% probability difference	N = 5VERY SMALL SAMPLE	3-2-0	60.0%	+0.8	15.4%	+7.3	0.00 (N 5)
8%+ probability difference	N = 8VERY SMALL SAMPLE	5-3-0	62.5%	+1.6	19.9%	+11.5	0.38 (N 8)

Total · line at projection time

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 29SMALL SAMPLE	14-15-0	48.3%	-1.6	-5.5%	+4.6	0.05 (N 29)
0-2 pts difference	N = 23SMALL SAMPLE	10-13-0	43.5%	-3.4	-14.7%	+3.4	0.07 (N 23)
2-4 pts difference	N = 5VERY SMALL SAMPLE	3-2-0	60.0%	+0.9	17.2%	+8.6	0.00 (N 5)
4-6 pts difference	N = 1VERY SMALL SAMPLE	1-0-0	100.0%	+0.9	92.6%	+12.0	0.00 (N 1)
0-2% probability difference	N = 6VERY SMALL SAMPLE	3-3-0	50.0%	-0.1	-2.4%	+1.0	0.17 (N 6)
2-4% probability difference	N = 8VERY SMALL SAMPLE	3-5-0	37.5%	-2.2	-27.2%	+3.1	0.13 (N 8)
4-6% probability difference	N = 9VERY SMALL SAMPLE	4-5-0	44.4%	-1.1	-11.8%	+5.2	-0.06 (N 9)
6-8% probability difference	N = 2VERY SMALL SAMPLE	1-1-0	50.0%	-0.1	-4.5%	+6.7	0.00 (N 2)
8%+ probability difference	N = 4VERY SMALL SAMPLE	3-1-0	75.0%	+1.9	46.9%	+10.4	0.00 (N 4)

ACTUAL TRACKED BETS

Only bets you enter manually. Nothing is added automatically from a model signal.

No bets entered (N = 0).

RETROSPECTIVE / DEVELOPMENT RESULTS

Backtests on games that were used to build, tune or choose these models (2022 to 2026 Week 6). Useful for comparing versions; NOT a track record and not mixed into the prospective results above. Win-probability scoring is the same for every version and both sections: see the note below.

game-projection-v1

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
2022–24 (train)	N = 815	0.4346	0.6362	66.4%	9.62	10.21	7.14	9.42 (model 9.62 on same 815)	10.12
2025 (validation)	N = 272	0.4450	0.6527	64.0%	10.05	10.61	7.44	9.72 (model 10.05 on same 272)	10.39
2026 W1–6 (test)	N = 92	0.4686	0.7009	57.6%	9.18	11.23	7.32	9.39 (model 9.18 on same 92)	10.59
All seasons · inexperienced-QB games	N = 326	0.4173	0.6094	69.6%	10.41	9.92	7.15	10.13 (model 10.41 on same 326)	9.67
All seasons · other games	N = 853	0.4481	0.6587	63.4%	9.41	10.55	7.25	9.24 (model 9.41 on same 853)	10.43

game-projection-v1.1

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
2022–24 (train)	N = 815	0.4363	0.6383	66.6%	9.67	10.19	7.16	9.42 (model 9.67 on same 815)	10.12
2025 (validation)	N = 272	0.4383	0.6454	64.3%	9.99	10.57	7.39	9.72 (model 9.99 on same 272)	10.39
2026 W1–6 (test)	N = 93	0.4541	0.6854	62.4%	9.04	11.31	7.28	9.31 (model 9.04 on same 93)	10.60
All seasons · inexperienced-QB games	N = 326	0.4181	0.6103	69.6%	10.41	9.90	7.16	10.13 (model 10.41 on same 326)	9.67
All seasons · other games	N = 854	0.4459	0.6564	64.3%	9.42	10.54	7.24	9.23 (model 9.42 on same 854)	10.43

game-projection-v1.2

Group	Games	Win-prob score: 3-outcome Brier (lower better)	3-outcome log loss	Winner picked	Margin MAE	Total MAE	Team-points MAE	Closing-line margin MAE	Closing total MAE
2022–24 (train)	N = 815	0.4351	0.6369	66.1%	9.64	10.15	7.14	9.42 (model 9.64 on same 815)	10.12
2025 (validation)	N = 272	0.4356	0.6425	64.7%	9.92	10.51	7.35	9.72 (model 9.92 on same 272)	10.39
2026 W1–6 (test)	N = 93	0.4527	0.6838	62.4%	8.99	11.23	7.24	9.31 (model 8.99 on same 93)	10.60
All seasons · inexperienced-QB games	N = 326	0.4124	0.6040	68.7%	10.27	9.76	7.08	10.13 (model 10.27 on same 326)	9.67
All seasons · other games	N = 854	0.4459	0.6564	64.3%	9.42	10.53	7.24	9.23 (model 9.42 on same 854)	10.43

Hypothetical Model Tracking against the closing line (retrospective)

Graded at nflverse closing lines, which were not available when a real projection would have been made. Development data only.

game-projection-v1

2022–24 (train)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	330-483-2	40.6%	-26.0	-3.2%	+7.4	—
0-2% probability difference	N = 128	64-64-0	50.0%	-8.9	-6.9%	+0.9	—
2-4% probability difference	N = 143	66-77-0	46.2%	+0.4	0.3%	+3.0	—
4-6% probability difference	N = 113	45-68-0	39.8%	-12.2	-10.8%	+5.0	—
6-8% probability difference	N = 111	34-76-1	30.9%	-26.5	-23.9%	+7.0	—
8%+ probability difference	N = 320	121-198-1	37.9%	+21.2	6.6%	+12.8	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	426-361-28	54.1%	+33.6	4.1%	+7.7	—
0-2 pts difference	N = 423	217-190-16	53.3%	+11.9	2.8%	+3.6	—
2-4 pts difference	N = 236	123-106-7	53.7%	+7.3	3.1%	+9.4	—
4-6 pts difference	N = 109	57-47-5	54.8%	+6.0	5.5%	+14.4	—
6+ pts difference	N = 47SMALL SAMPLE	29-18-0	61.7%	+8.4	17.9%	+20.1	—
0-2% probability difference	N = 136	72-57-7	55.8%	+9.6	7.0%	+1.0	—
2-4% probability difference	N = 134	71-56-7	55.9%	+9.4	7.0%	+3.0	—
4-6% probability difference	N = 104	55-48-1	53.4%	+2.6	2.5%	+5.0	—
6-8% probability difference	N = 111	49-60-2	45.0%	-14.6	-13.1%	+7.0	—
8%+ probability difference	N = 330	179-140-11	56.1%	+26.6	8.1%	+13.4	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	431-376-8	53.4%	+21.1	2.6%	+6.5	—
0-2 pts difference	N = 451	231-218-2	51.4%	-4.6	-1.0%	+3.0	—
2-4 pts difference	N = 254	138-112-4	55.2%	+15.1	5.9%	+8.7	—
4-6 pts difference	N = 85	48-37-0	56.5%	+6.8	8.0%	+14.4	—
6+ pts difference	N = 25SMALL SAMPLE	14-9-2	60.9%	+3.8	15.3%	+21.1	—
0-2% probability difference	N = 166	83-83-0	50.0%	-6.9	-4.1%	+1.1	—
2-4% probability difference	N = 140	72-67-1	51.8%	-0.4	-0.3%	+2.9	—
4-6% probability difference	N = 135	71-63-1	53.0%	+2.8	2.1%	+4.9	—
6-8% probability difference	N = 108	62-44-2	58.5%	+12.6	11.6%	+6.9	—
8%+ probability difference	N = 266	143-119-4	54.6%	+12.9	4.8%	+12.4	—

2025 (validation)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	99-172-1	36.5%	-19.0	-7.0%	+7.8	—
0-2% probability difference	N = 38SMALL SAMPLE	19-19-0	50.0%	+1.0	2.6%	+1.0	—
2-4% probability difference	N = 33SMALL SAMPLE	13-20-0	39.4%	-8.0	-24.2%	+3.0	—
4-6% probability difference	N = 42SMALL SAMPLE	16-26-0	38.1%	-4.8	-11.4%	+4.9	—
6-8% probability difference	N = 40SMALL SAMPLE	13-27-0	32.5%	-4.6	-11.4%	+6.9	—
8%+ probability difference	N = 119	38-80-1	32.2%	-2.6	-2.2%	+12.6	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	141-130-1	52.0%	-0.7	-0.3%	+8.1	—
0-2 pts difference	N = 136	71-65-0	52.2%	+0.3	0.2%	+4.0	—
2-4 pts difference	N = 84	43-41-0	51.2%	-1.7	-2.0%	+9.8	—
4-6 pts difference	N = 39SMALL SAMPLE	21-18-0	53.8%	+1.3	3.3%	+14.6	—
6+ pts difference	N = 13VERY SMALL SAMPLE	6-6-1	50.0%	-0.5	-4.2%	+22.0	—
0-2% probability difference	N = 39SMALL SAMPLE	24-15-0	61.5%	+7.0	18.0%	+0.9	—
2-4% probability difference	N = 34SMALL SAMPLE	14-20-0	41.2%	-7.1	-20.9%	+3.0	—
4-6% probability difference	N = 46SMALL SAMPLE	24-22-0	52.2%	-0.3	-0.6%	+5.0	—
6-8% probability difference	N = 28SMALL SAMPLE	16-12-0	57.1%	+2.5	8.8%	+6.9	—
8%+ probability difference	N = 125	63-61-1	50.8%	-2.8	-2.2%	+13.2	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	136-136-0	50.0%	-11.7	-4.3%	+6.3	—
0-2 pts difference	N = 156	79-77-0	50.6%	-5.0	-3.2%	+3.1	—
2-4 pts difference	N = 78	40-38-0	51.3%	-1.2	-1.6%	+8.3	—
4-6 pts difference	N = 30SMALL SAMPLE	10-20-0	33.3%	-10.8	-35.9%	+13.5	—
6+ pts difference	N = 8VERY SMALL SAMPLE	7-1-0	87.5%	+5.3	66.1%	+20.0	—
0-2% probability difference	N = 51	24-27-0	47.1%	-5.1	-10.0%	+1.2	—
2-4% probability difference	N = 56	34-22-0	60.7%	+8.7	15.5%	+2.9	—
4-6% probability difference	N = 45SMALL SAMPLE	22-23-0	48.9%	-3.0	-6.8%	+4.9	—
6-8% probability difference	N = 43SMALL SAMPLE	17-26-0	39.5%	-10.3	-24.0%	+7.0	—
8%+ probability difference	N = 77	39-38-0	50.6%	-1.9	-2.4%	+12.4	—

2026 W1–6 (test)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 92	36-55-1	39.6%	+6.2	6.7%	+8.2	—
0-2% probability difference	N = 12VERY SMALL SAMPLE	3-8-1	27.3%	-5.7	-47.2%	+1.0	—
2-4% probability difference	N = 11VERY SMALL SAMPLE	6-5-0	54.5%	+2.7	24.9%	+2.9	—
4-6% probability difference	N = 17VERY SMALL SAMPLE	6-11-0	35.3%	-3.6	-21.4%	+5.3	—
6-8% probability difference	N = 8VERY SMALL SAMPLE	3-5-0	37.5%	-0.2	-2.8%	+7.0	—
8%+ probability difference	N = 44SMALL SAMPLE	18-26-0	40.9%	+13.0	29.5%	+12.8	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 92	47-39-6	54.7%	+4.2	4.6%	+8.6	—
0-2 pts difference	N = 46SMALL SAMPLE	24-19-3	55.8%	+3.1	6.8%	+4.2	—
2-4 pts difference	N = 26SMALL SAMPLE	11-12-3	47.8%	-1.9	-7.2%	+10.5	—
4-6 pts difference	N = 16VERY SMALL SAMPLE	9-7-0	56.3%	+1.3	8.1%	+15.0	—
6+ pts difference	N = 4VERY SMALL SAMPLE	3-1-0	75.0%	+1.6	41.2%	+20.4	—
0-2% probability difference	N = 10VERY SMALL SAMPLE	7-3-0	70.0%	+3.4	33.7%	+1.3	—
2-4% probability difference	N = 10VERY SMALL SAMPLE	5-4-1	55.6%	+0.4	4.1%	+2.8	—
4-6% probability difference	N = 18VERY SMALL SAMPLE	7-10-1	41.2%	-3.4	-18.6%	+4.9	—
6-8% probability difference	N = 11VERY SMALL SAMPLE	3-7-1	30.0%	-4.4	-39.6%	+7.3	—
8%+ probability difference	N = 43SMALL SAMPLE	25-15-3	62.5%	+8.1	18.9%	+13.5	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 92	39-53-0	42.4%	-17.1	-18.5%	+7.2	—
0-2 pts difference	N = 42SMALL SAMPLE	16-26-0	38.1%	-11.2	-26.6%	+3.3	—
2-4 pts difference	N = 36SMALL SAMPLE	17-19-0	47.2%	-3.3	-9.2%	+8.6	—
4-6 pts difference	N = 11VERY SMALL SAMPLE	4-7-0	36.4%	-3.3	-30.3%	+14.8	—
6+ pts difference	N = 3VERY SMALL SAMPLE	2-1-0	66.7%	+0.8	25.4%	+17.0	—
0-2% probability difference	N = 14VERY SMALL SAMPLE	6-8-0	42.9%	-2.6	-18.5%	+1.2	—
2-4% probability difference	N = 13VERY SMALL SAMPLE	7-6-0	53.8%	+0.7	5.2%	+2.9	—
4-6% probability difference	N = 13VERY SMALL SAMPLE	5-8-0	38.5%	-3.5	-26.8%	+4.8	—
6-8% probability difference	N = 15VERY SMALL SAMPLE	5-10-0	33.3%	-5.4	-36.1%	+7.3	—
8%+ probability difference	N = 37SMALL SAMPLE	16-21-0	43.2%	-6.2	-16.8%	+11.8	—

game-projection-v1.1

2022–24 (train)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	328-485-2	40.3%	-55.1	-6.8%	+6.9	—
0-2% probability difference	N = 155	64-91-0	41.3%	-35.0	-22.6%	+1.0	—
2-4% probability difference	N = 133	60-72-1	45.5%	-10.3	-7.7%	+3.0	—
4-6% probability difference	N = 137	60-77-0	43.8%	+9.3	6.8%	+4.9	—
6-8% probability difference	N = 108	45-63-0	41.7%	-10.4	-9.6%	+7.0	—
8%+ probability difference	N = 282	99-182-1	35.2%	-8.8	-3.1%	+12.9	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	416-371-28	52.9%	+13.5	1.7%	+7.2	—
0-2 pts difference	N = 454	235-204-15	53.5%	+13.8	3.0%	+3.4	—
2-4 pts difference	N = 227	108-110-9	49.5%	-10.3	-4.5%	+9.7	—
4-6 pts difference	N = 100	52-44-4	54.2%	+4.1	4.1%	+14.3	—
6+ pts difference	N = 34SMALL SAMPLE	21-13-0	61.8%	+6.0	17.5%	+20.4	—
0-2% probability difference	N = 157	72-77-8	48.3%	-10.9	-7.0%	+1.0	—
2-4% probability difference	N = 136	76-56-4	57.6%	+14.1	10.4%	+3.1	—
4-6% probability difference	N = 116	66-48-2	57.9%	+12.9	11.1%	+5.0	—
6-8% probability difference	N = 102	49-50-3	49.5%	-4.4	-4.3%	+7.0	—
8%+ probability difference	N = 304	153-140-11	52.2%	+1.9	0.6%	+13.2	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	420-387-8	52.0%	+0.4	0.1%	+6.1	—
0-2 pts difference	N = 475	239-235-1	50.4%	-13.8	-2.9%	+2.9	—
2-4 pts difference	N = 246	128-113-5	53.1%	+4.7	1.9%	+8.7	—
4-6 pts difference	N = 72	40-31-1	56.3%	+5.8	8.1%	+14.2	—
6+ pts difference	N = 22SMALL SAMPLE	13-8-1	61.9%	+3.8	17.2%	+20.4	—
0-2% probability difference	N = 173	81-92-0	46.8%	-17.2	-10.0%	+1.0	—
2-4% probability difference	N = 163	73-90-0	44.8%	-22.8	-14.0%	+3.0	—
4-6% probability difference	N = 128	75-52-1	59.1%	+17.2	13.4%	+5.0	—
6-8% probability difference	N = 105	56-46-3	54.9%	+5.7	5.4%	+6.9	—
8%+ probability difference	N = 246	135-107-4	55.8%	+17.5	7.1%	+12.1	—

2025 (validation)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	100-171-1	36.9%	-22.6	-8.3%	+6.5	—
0-2% probability difference	N = 44SMALL SAMPLE	20-24-0	45.5%	-8.2	-18.6%	+0.9	—
2-4% probability difference	N = 51	19-32-0	37.3%	-11.6	-22.7%	+3.1	—
4-6% probability difference	N = 51	20-30-1	40.0%	+8.1	15.9%	+4.9	—
6-8% probability difference	N = 38SMALL SAMPLE	11-27-0	28.9%	-7.0	-18.4%	+7.0	—
8%+ probability difference	N = 88	30-58-0	34.1%	-3.9	-4.5%	+12.1	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	135-136-1	49.8%	-12.1	-4.4%	+6.9	—
0-2 pts difference	N = 165	79-86-0	47.9%	-13.7	-8.3%	+3.9	—
2-4 pts difference	N = 70	38-32-0	54.3%	+3.2	4.5%	+9.5	—
4-6 pts difference	N = 28SMALL SAMPLE	14-14-0	50.0%	-1.2	-4.3%	+13.4	—
6+ pts difference	N = 9VERY SMALL SAMPLE	4-4-1	50.0%	-0.4	-4.0%	+21.4	—
0-2% probability difference	N = 45SMALL SAMPLE	26-19-0	57.8%	+4.8	10.6%	+1.0	—
2-4% probability difference	N = 47SMALL SAMPLE	20-27-0	42.6%	-8.9	-19.0%	+3.0	—
4-6% probability difference	N = 49SMALL SAMPLE	25-24-0	51.0%	-1.2	-2.4%	+4.8	—
6-8% probability difference	N = 32SMALL SAMPLE	13-19-0	40.6%	-7.1	-22.2%	+6.9	—
8%+ probability difference	N = 99	51-47-1	52.0%	+0.3	0.3%	+12.5	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	130-142-0	47.8%	-23.0	-8.5%	+5.9	—
0-2 pts difference	N = 158	71-87-0	44.9%	-22.0	-13.9%	+2.9	—
2-4 pts difference	N = 87	44-43-0	50.6%	-2.9	-3.3%	+8.5	—
4-6 pts difference	N = 23SMALL SAMPLE	12-11-0	52.2%	+0.1	0.4%	+14.3	—
6+ pts difference	N = 4VERY SMALL SAMPLE	3-1-0	75.0%	+1.7	41.8%	+18.7	—
0-2% probability difference	N = 55	21-34-0	38.2%	-15.1	-27.4%	+0.9	—
2-4% probability difference	N = 56	29-27-0	51.8%	-0.6	-1.1%	+2.9	—
4-6% probability difference	N = 51	21-30-0	41.2%	-10.7	-21.1%	+5.1	—
6-8% probability difference	N = 32SMALL SAMPLE	21-11-0	65.6%	+8.3	25.8%	+6.9	—
8%+ probability difference	N = 78	38-40-0	48.7%	-4.9	-6.2%	+11.7	—

2026 W1–6 (test)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	39-53-1	42.4%	+10.9	11.7%	+7.3	—
0-2% probability difference	N = 15VERY SMALL SAMPLE	6-9-0	40.0%	-3.8	-25.5%	+1.1	—
2-4% probability difference	N = 14VERY SMALL SAMPLE	2-12-0	14.3%	-11.2	-79.8%	+3.2	—
4-6% probability difference	N = 11VERY SMALL SAMPLE	6-4-1	60.0%	+7.4	66.9%	+5.1	—
6-8% probability difference	N = 21SMALL SAMPLE	8-13-0	38.1%	-1.7	-8.1%	+7.0	—
8%+ probability difference	N = 32SMALL SAMPLE	17-15-0	53.1%	+20.2	63.2%	+12.9	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	48-39-6	55.2%	+4.9	5.3%	+7.5	—
0-2 pts difference	N = 56	26-24-6	52.0%	-0.5	-0.8%	+4.6	—
2-4 pts difference	N = 23SMALL SAMPLE	13-10-0	56.5%	+2.2	9.5%	+10.3	—
4-6 pts difference	N = 12VERY SMALL SAMPLE	7-5-0	58.3%	+1.4	11.4%	+13.9	—
6+ pts difference	N = 2VERY SMALL SAMPLE	2-0-0	100.0%	+1.8	91.8%	+20.7	—
0-2% probability difference	N = 8VERY SMALL SAMPLE	5-3-0	62.5%	+1.3	16.6%	+0.9	—
2-4% probability difference	N = 17VERY SMALL SAMPLE	8-8-1	50.0%	-0.7	-4.4%	+2.9	—
4-6% probability difference	N = 15VERY SMALL SAMPLE	6-7-2	46.2%	-1.7	-11.5%	+4.9	—
6-8% probability difference	N = 17VERY SMALL SAMPLE	7-8-2	46.7%	-1.3	-7.8%	+7.0	—
8%+ probability difference	N = 36SMALL SAMPLE	22-13-1	62.9%	+7.4	20.5%	+12.5	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	37-56-0	39.8%	-22.0	-23.6%	+6.8	—
0-2 pts difference	N = 44SMALL SAMPLE	20-24-0	45.5%	-5.4	-12.2%	+3.4	—
2-4 pts difference	N = 39SMALL SAMPLE	12-27-0	30.8%	-16.2	-41.5%	+8.3	—
4-6 pts difference	N = 7VERY SMALL SAMPLE	3-4-0	42.9%	-1.2	-17.4%	+14.7	—
6+ pts difference	N = 3VERY SMALL SAMPLE	2-1-0	66.7%	+0.8	27.4%	+18.6	—
0-2% probability difference	N = 15VERY SMALL SAMPLE	7-8-0	46.7%	-1.5	-9.7%	+1.2	—
2-4% probability difference	N = 13VERY SMALL SAMPLE	5-8-0	38.5%	-3.4	-26.2%	+3.0	—
4-6% probability difference	N = 13VERY SMALL SAMPLE	5-8-0	38.5%	-3.4	-26.5%	+5.1	—
6-8% probability difference	N = 19VERY SMALL SAMPLE	6-13-0	31.6%	-7.4	-38.8%	+6.9	—
8%+ probability difference	N = 33SMALL SAMPLE	14-19-0	42.4%	-6.3	-19.0%	+11.4	—

game-projection-v1.2

2022–24 (train)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	343-470-2	42.2%	-33.6	-4.1%	+6.7	—
0-2% probability difference	N = 159	75-84-0	47.2%	-16.0	-10.1%	+0.9	—
2-4% probability difference	N = 137	65-71-1	47.8%	-0.8	-0.6%	+2.9	—
4-6% probability difference	N = 136	58-78-0	42.6%	-4.0	-2.9%	+4.9	—
6-8% probability difference	N = 115	48-67-0	41.7%	-2.7	-2.4%	+7.0	—
8%+ probability difference	N = 268	97-170-1	36.3%	-10.1	-3.8%	+12.9	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	425-362-28	54.0%	+30.9	3.8%	+7.1	—
0-2 pts difference	N = 460	243-201-16	54.7%	+24.2	5.3%	+3.5	—
2-4 pts difference	N = 226	112-106-8	51.4%	-2.7	-1.2%	+9.4	—
4-6 pts difference	N = 97	50-43-4	53.8%	+3.4	3.5%	+14.3	—
6+ pts difference	N = 32SMALL SAMPLE	20-12-0	62.5%	+6.0	18.6%	+20.6	—
0-2% probability difference	N = 151	82-61-8	57.3%	+14.5	9.6%	+1.0	—
2-4% probability difference	N = 139	74-60-5	55.2%	+8.3	6.0%	+3.1	—
4-6% probability difference	N = 128	70-55-3	56.0%	+9.3	7.3%	+4.9	—
6-8% probability difference	N = 97	46-49-2	48.4%	-6.0	-6.2%	+6.9	—
8%+ probability difference	N = 300	153-137-10	52.8%	+4.8	1.6%	+13.0	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 815	423-384-8	52.4%	+6.2	0.8%	+5.9	—
0-2 pts difference	N = 485	246-238-1	50.8%	-10.1	-2.1%	+2.9	—
2-4 pts difference	N = 248	131-111-6	54.1%	+9.3	3.8%	+8.6	—
4-6 pts difference	N = 62	35-27-0	56.5%	+5.0	8.1%	+14.6	—
6+ pts difference	N = 20SMALL SAMPLE	11-8-1	57.9%	+2.0	9.9%	+20.1	—
0-2% probability difference	N = 185	83-102-0	44.9%	-25.2	-13.6%	+0.9	—
2-4% probability difference	N = 163	80-83-0	49.1%	-9.4	-5.8%	+3.0	—
4-6% probability difference	N = 124	76-47-1	61.8%	+22.9	18.5%	+5.1	—
6-8% probability difference	N = 110	54-53-3	50.5%	-3.0	-2.8%	+6.9	—
8%+ probability difference	N = 233	130-99-4	56.8%	+20.9	9.0%	+12.0	—

2025 (validation)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	99-172-1	36.5%	-27.4	-10.1%	+6.4	—
0-2% probability difference	N = 41SMALL SAMPLE	14-27-0	34.1%	-16.2	-39.5%	+0.8	—
2-4% probability difference	N = 58	22-36-0	37.9%	-12.5	-21.5%	+3.2	—
4-6% probability difference	N = 52	19-32-1	37.3%	+3.9	7.5%	+4.9	—
6-8% probability difference	N = 38SMALL SAMPLE	17-21-0	44.7%	+8.7	23.0%	+7.0	—
8%+ probability difference	N = 83	27-56-0	32.5%	-11.4	-13.7%	+12.2	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	134-137-1	49.4%	-13.9	-5.1%	+6.8	—
0-2 pts difference	N = 170	80-90-0	47.1%	-16.5	-9.7%	+4.0	—
2-4 pts difference	N = 67	35-32-0	52.2%	+0.2	0.2%	+9.5	—
4-6 pts difference	N = 27SMALL SAMPLE	16-11-0	59.3%	+3.6	13.3%	+13.8	—
6+ pts difference	N = 8VERY SMALL SAMPLE	3-4-1	42.9%	-1.2	-15.4%	+21.0	—
0-2% probability difference	N = 44SMALL SAMPLE	22-22-0	50.0%	-1.8	-4.0%	+1.0	—
2-4% probability difference	N = 49SMALL SAMPLE	21-28-0	42.9%	-8.9	-18.2%	+3.1	—
4-6% probability difference	N = 49SMALL SAMPLE	24-25-0	49.0%	-3.1	-6.3%	+4.8	—
6-8% probability difference	N = 36SMALL SAMPLE	20-16-0	55.6%	+2.4	6.7%	+6.9	—
8%+ probability difference	N = 94	47-46-1	50.5%	-2.6	-2.7%	+12.5	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 272	133-139-0	48.9%	-17.2	-6.3%	+5.9	—
0-2 pts difference	N = 158	70-88-0	44.3%	-23.7	-15.0%	+2.9	—
2-4 pts difference	N = 85	47-38-0	55.3%	+4.8	5.6%	+8.3	—
4-6 pts difference	N = 24SMALL SAMPLE	13-11-0	54.2%	+1.1	4.5%	+13.8	—
6+ pts difference	N = 5VERY SMALL SAMPLE	3-2-0	60.0%	+0.7	13.4%	+18.7	—
0-2% probability difference	N = 55	22-33-0	40.0%	-13.0	-23.7%	+1.0	—
2-4% probability difference	N = 59	29-30-0	49.2%	-3.4	-5.7%	+3.0	—
4-6% probability difference	N = 46SMALL SAMPLE	21-25-0	45.7%	-5.8	-12.6%	+5.1	—
6-8% probability difference	N = 39SMALL SAMPLE	25-14-0	64.1%	+8.7	22.3%	+7.0	—
8%+ probability difference	N = 73	36-37-0	49.3%	-3.7	-5.0%	+11.8	—

2026 W1–6 (test)

Moneyline · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	39-53-1	42.4%	+10.9	11.7%	+7.1	—
0-2% probability difference	N = 19VERY SMALL SAMPLE	7-12-0	36.8%	-6.3	-33.2%	+1.0	—
2-4% probability difference	N = 11VERY SMALL SAMPLE	1-10-0	9.1%	-9.7	-88.1%	+2.9	—
4-6% probability difference	N = 10VERY SMALL SAMPLE	6-3-1	66.7%	+9.6	95.6%	+5.2	—
6-8% probability difference	N = 21SMALL SAMPLE	7-14-0	33.3%	-4.5	-21.6%	+6.9	—
8%+ probability difference	N = 32SMALL SAMPLE	18-14-0	56.3%	+21.8	68.2%	+12.8	—

Spread · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	50-37-6	57.5%	+8.8	9.5%	+7.3	—
0-2 pts difference	N = 55	26-23-6	53.1%	+0.7	1.3%	+4.2	—
2-4 pts difference	N = 24SMALL SAMPLE	15-9-0	62.5%	+4.9	20.4%	+10.2	—
4-6 pts difference	N = 12VERY SMALL SAMPLE	7-5-0	58.3%	+1.4	11.4%	+13.7	—
6+ pts difference	N = 2VERY SMALL SAMPLE	2-0-0	100.0%	+1.8	91.8%	+20.7	—
0-2% probability difference	N = 12VERY SMALL SAMPLE	8-4-0	66.7%	+3.1	26.0%	+0.8	—
2-4% probability difference	N = 16VERY SMALL SAMPLE	8-6-2	57.1%	+1.3	7.9%	+2.9	—
4-6% probability difference	N = 12VERY SMALL SAMPLE	4-7-1	36.4%	-3.5	-29.2%	+4.8	—
6-8% probability difference	N = 17VERY SMALL SAMPLE	7-8-2	46.7%	-1.3	-7.9%	+7.1	—
8%+ probability difference	N = 36SMALL SAMPLE	23-12-1	65.7%	+9.3	25.7%	+12.5	—

Total · closing line

Bucket	Signals	W-L-P	Win %	Hypothetical units	Hypothetical ROI	Avg probability difference	Avg CLV
All	N = 93	38-55-0	40.9%	-20.1	-21.6%	+6.6	—
0-2 pts difference	N = 46SMALL SAMPLE	21-25-0	45.7%	-5.5	-12.0%	+3.4	—
2-4 pts difference	N = 38SMALL SAMPLE	12-26-0	31.6%	-15.1	-39.8%	+8.3	—
4-6 pts difference	N = 7VERY SMALL SAMPLE	3-4-0	42.9%	-1.2	-17.4%	+15.1	—
6+ pts difference	N = 2VERY SMALL SAMPLE	2-0-0	100.0%	+1.8	91.1%	+18.3	—
0-2% probability difference	N = 15VERY SMALL SAMPLE	7-8-0	46.7%	-1.5	-9.7%	+1.3	—
2-4% probability difference	N = 14VERY SMALL SAMPLE	6-8-0	42.9%	-2.5	-17.8%	+2.9	—
4-6% probability difference	N = 15VERY SMALL SAMPLE	6-9-0	40.0%	-3.5	-23.5%	+5.2	—
6-8% probability difference	N = 20SMALL SAMPLE	6-14-0	30.0%	-8.5	-42.5%	+6.9	—
8%+ probability difference	N = 29SMALL SAMPLE	13-16-0	44.8%	-4.1	-14.1%	+11.6	—

How win probability is scored

The simulator plays overtime under the season's rules, so every projection has a final-game probability for HOME WIN, TIE and AWAY WIN. Win-probability accuracy is scored with the 3-outcome Brier score: (P home − home won)² + (P tie − tie)² + (P away − away won)², where 0 is perfect and 2 is worst, with ties included. The 3-outcome log loss is shown alongside it. The 3-outcome Brier is roughly double the size of a home/away-only Brier, so the two can't be compared directly. "Winner picked" counts a tie as a miss. Comparisons with the sportsbook moneyline use the probability excluding ties, because moneylines refund ties. One method for all versions and all results since 2026-10-21.