Insights · NFL

What happened. What we learned.

Every graded pick, wins and losses, in one place. Read the record beside the proof, then see which angles held up.

Leaderboard · hit rate by lens

Leaderboard · models against the favourite baseline

Headline performance is ATS spread win rate. Pushes remain in graded spread plays but are excluded from the ATS spread denominator.

Rows are ordered by spread win rate; closing-line value breaks ties when available.
Entry	Kind	Graded spread plays	ATS spread win rate	Record (W-L-P)	CLV · secondary
[Auto] NFL Spread (MARS) n=50 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	50	64.0% · n=50 Live picks · all time · n=50 · through Oct 18, 1:08 PM ET	32-18-0	−0.020
[Auto] NFL Spread (MARS) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	60.0% · n=60 Live picks · all time · n=60 · through Oct 18, 1:08 PM ET	36-24-21	+0.451
[Auto] Blueprint (Py): Momentum Spread Model (ElasticNet) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	57.4% · n=68 Live picks · all time · n=68 · through Oct 18, 1:08 PM ET	39-29-13	+0.074
[Auto] NFL Elo Power Ratings Spread (XGBoost) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	56.5% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	35-27-0	+0.024
[Auto] NFL Spread (SVR) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	55.7% · n=61 Live picks · all time · n=61 · through Oct 18, 1:08 PM ET	34-27-20	+0.389
[Auto] NFL Spread (ExtraTrees) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	55.4% · n=56 Live picks · all time · n=56 · through Oct 18, 1:08 PM ET	31-25-25	+0.117
[Auto] NFL Elo Power Ratings Spread (XGBoost) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	54.5% · n=66 Live picks · all time · n=66 · through Oct 18, 1:08 PM ET	36-30-15	+0.117
[Auto] NFL Spread (KNN) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	54.4% · n=57 Live picks · all time · n=57 · through Oct 18, 1:08 PM ET	31-26-24	+0.216
model-3950f12e n=24 graded spread plays · as of 2026-09-22T03:34:00.49759+00:00	model	24	54.2% · n=24 Live picks · all time · n=24 · through Sep 21, 11:34 PM ET	13-11-0	+0.250
[Auto] Blueprint (Py): Momentum Spread Model (XGB) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	54.0% · n=63 Live picks · all time · n=63 · through Oct 18, 1:08 PM ET	34-29-18	−0.025
[Auto] Blueprint (Py): Full Feature Spread (Stacking Ensemble) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	53.6% · n=56 Live picks · all time · n=56 · through Oct 18, 1:08 PM ET	30-26-25	+0.481
[Auto] Blueprint (Py): Turnover Regression Model (ElasticNet) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	52.6% · n=57 Live picks · all time · n=57 · through Oct 18, 1:08 PM ET	30-27-24	+0.302
[Auto] Playbook (Py): Full Feature Spread (Stacking Ensemble) n=61 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	61	52.5% · n=61 Live picks · all time · n=61 · through Oct 18, 1:08 PM ET	32-29-0	+0.016
[Auto] NFL ATS XGBoost (Custom Objective) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	52.4% · n=63 Live picks · all time · n=63 · through Oct 18, 1:08 PM ET	33-30-18	+0.302
[Auto] NFL Spread (HistGradientBoosting) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	51.7% · n=58 Live picks · all time · n=58 · through Oct 18, 1:08 PM ET	30-28-23	+0.142
[Auto] NFL Spread (KNN) n=60 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	60	51.7% · n=60 Live picks · all time · n=60 · through Oct 18, 1:08 PM ET	31-29-0	−0.042
[Auto] NFL GNN Matchup n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	51.6% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	32-30-0	+0.105
[Auto] NFL Spread Stacking n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	51.6% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	32-30-0	+0.065
[Auto] NFL GNN Matchup n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	50.9% · n=57 Live picks · all time · n=57 · through Oct 18, 1:08 PM ET	29-28-24	+0.475
[Auto] Playbook (Py): Full Feature Spread (Mixture of Experts) n=61 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	61	50.8% · n=61 Live picks · all time · n=61 · through Oct 18, 1:08 PM ET	31-30-0	+0.008
[Auto] NFL Spread Stacking n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	50.8% · n=65 Live picks · all time · n=65 · through Oct 18, 1:08 PM ET	33-32-16	+0.389
[Auto] Blueprint (Py): Full Feature Spread (XGB) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	50.0% · n=64 Live picks · all time · n=64 · through Oct 18, 1:08 PM ET	32-32-17	+0.259
model-4cbee016 n=42 graded spread plays · as of 2026-10-16T04:08:00.098049+00:00	model	42	50.0% · n=38 Live picks · all time · n=38 · through Oct 16, 12:08 AM ET	19-19-4	+0.214
[Auto] Blueprint (Py): Full Feature Spread (Mixture of Experts) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	50.0% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	31-31-0	+0.089
[Auto] NFL Spread (SVR) n=78 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	78	50.0% · n=78 Live picks · all time · n=78 · through Oct 18, 1:08 PM ET	39-39-0	+0.013
[Auto] NFL ATS XGBoost (Custom Objective) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	50.0% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	31-31-0	0.000
[Auto] NFL Spread (AdaBoost) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	49.2% · n=59 Live picks · all time · n=59 · through Oct 18, 1:08 PM ET	29-30-22	+0.216
[Auto] Blueprint (Py): Pressure & Efficiency Model (LightGBM) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	48.4% · n=64 Live picks · all time · n=64 · through Oct 18, 1:08 PM ET	31-33-17	+0.117
[Auto] Blueprint (Py): Full Feature Spread (Mixture of Experts) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	48.2% · n=56 Live picks · all time · n=56 · through Oct 18, 1:08 PM ET	27-29-25	+0.241
[Auto] Playbook (Py): Feature Store Spread Model (XGB) n=46 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	46	47.8% · n=46 Live picks · all time · n=46 · through Oct 18, 1:08 PM ET	22-24-0	+0.022
[Auto] Playbook (Py): Full Feature Spread (XGB) n=46 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	46	47.8% · n=46 Live picks · all time · n=46 · through Oct 18, 1:08 PM ET	22-24-0	+0.022
[Auto] Playbook (Py): Momentum Spread Model (XGB) n=46 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	46	47.8% · n=46 Live picks · all time · n=46 · through Oct 18, 1:08 PM ET	22-24-0	+0.022
[Auto] Blueprint (Py): Full Feature Spread (XGB) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	47.5% · n=61 Live picks · all time · n=61 · through Oct 18, 1:08 PM ET	29-32-20	+0.327
[Auto] NFL Minimal Signal Spread (XGBoost) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	47.5% · n=61 Live picks · all time · n=61 · through Oct 18, 1:08 PM ET	29-32-20	+0.142
[Auto] NFL MLP Spread n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	47.5% · n=59 Live picks · all time · n=59 · through Oct 18, 1:08 PM ET	28-31-22	+0.327
[Auto] NFL MLP Spread n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	46.8% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	29-33-0	+0.008
[Auto] NFL Spread (Gaussian Process) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	46.8% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	29-33-0	+0.008
model-1758a821 n=30 graded spread plays · as of 2026-09-23T05:56:07.815427+00:00	model	30	46.7% · n=30 Live picks · all time · n=30 · through Sep 23, 1:56 AM ET	14-16-0	+0.200
[Auto] Blueprint (Py): Momentum Spread Model (ElasticNet) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	46.3% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	37-43-1	+0.062
[Auto] NFL Market Efficiency Spread (XGBoost) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.8% · n=59 Live picks · all time · n=59 · through Oct 18, 1:08 PM ET	27-32-22	+0.142
[Auto] NFL Multi-Task Spread n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.3% · n=64 Live picks · all time · n=64 · through Oct 18, 1:08 PM ET	29-35-17	+0.438
[Auto] NFL Spread (Gaussian Process) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.2% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	28-34-19	+0.364
[Auto] NFL Spread (CatBoost) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	45.2% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	28-34-0	−0.065
[Auto] Blueprint (Py): Turnover Regression Model (ElasticNet) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.0% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	36-44-1	+0.296
[Auto] Blueprint (Py): Turnover Regression Model (ElasticNet) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.0% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	36-44-1	+0.247
[Auto] Blueprint (Py): Pressure & Efficiency Model (LightGBM) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	45.0% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	36-44-1	+0.160
[Auto] Blueprint (Py): Full Feature Spread (Stacking Ensemble) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	44.8% · n=58 Live picks · all time · n=58 · through Oct 18, 1:08 PM ET	26-32-23	+0.389
[Auto] Blueprint (Py): SHAP Explainable Elo Model (Linear) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	44.6% · n=65 Live picks · all time · n=65 · through Oct 18, 1:08 PM ET	29-36-16	+0.136
[Auto] Playbook (Py): Momentum Spread Model (ElasticNet) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	43.5% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	27-35-0	+0.040
[Auto] NFL Spread (ExtraTrees) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	43.5% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	27-35-0	0.000
[Auto] Blueprint (Py): Simple Elo Spread Model (Linear) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	42.6% · n=68 Live picks · all time · n=68 · through Oct 18, 1:08 PM ET	29-39-13	+0.099
[Auto] NFL Multi-Task Spread n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	41.9% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	26-36-0	+0.065
[Auto] NFL Spread (HistGradientBoosting) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	41.9% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	26-36-0	+0.024
[Auto] Blueprint (Py): Simple Elo Spread Model (Bayesian) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	40.7% · n=59 Live picks · all time · n=59 · through Oct 18, 1:08 PM ET	24-35-22	+0.136
[Auto] Playbook (Py): Pressure & Efficiency Model (LightGBM) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	40.3% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	25-37-0	−0.065
[Auto] Blueprint (Py): SHAP Explainable Elo Model (Linear) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	40.0% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	32-48-1	+0.136
[Auto] Blueprint (Py): Simple Elo Spread Model (Bayesian) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	39.7% · n=58 Live picks · all time · n=58 · through Oct 18, 1:08 PM ET	23-35-23	+0.179
[Auto] Blueprint (Py): Simple Elo Spread Model (Linear) n=81 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	81	38.8% · n=80 Live picks · all time · n=80 · through Oct 18, 1:08 PM ET	31-49-1	+0.099
[Auto] Playbook (Py): Turnover Regression Model (ElasticNet) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	38.7% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	24-38-0	+0.040
[Auto] NFL Spread (AdaBoost) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	38.7% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	24-38-0	+0.024
[Auto] Blueprint (Py): Pressure & Efficiency Model (LightGBM) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	38.7% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	24-38-0	−0.073
[Auto] Playbook (Py): Simple Elo Spread Model (Bayesian) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	37.1% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	23-39-0	−0.097
[Auto] Playbook (Py): SHAP Explainable Elo Model (Linear) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	37.1% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	23-39-0	−0.105
[Auto] Sharkai Starter: NFL Spread n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	35.5% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	22-40-0	−0.024
[Auto] Playbook (Py): Simple Elo Spread Model (Linear) n=62 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	62	35.5% · n=62 Live picks · all time · n=62 · through Oct 18, 1:08 PM ET	22-40-0	−0.065
[Auto] NFL Market Efficiency Spread (XGBoost) n=46 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	46	34.8% · n=46 Live picks · all time · n=46 · through Oct 18, 1:08 PM ET	16-30-0	−0.152
[Auto] NFL Minimal Signal Spread (XGBoost) n=46 graded spread plays · as of 2026-10-18T17:08:00.090952+00:00	model	46	32.6% · n=46 Live picks · all time · n=46 · through Oct 18, 1:08 PM ET	15-31-0	−0.033
[Auto] MLB Run Line (Gaussian Process) n=115 graded spread plays · as of 2026-05-20T09:00:00.606554+00:00	model	115	—	0-0-115	—
[Auto] MLB Run Line (HistGradientBoosting) n=115 graded spread plays · as of 2026-05-20T09:00:00.606554+00:00	model	115	—	0-0-115	—
[Auto] Soccer Spread (Gaussian Process) n=20 graded spread plays · as of 2026-09-02T02:48:08.572892+00:00	model	20	—	0-0-20	—
[Auto] Soccer Spread (HistGradientBoosting) n=20 graded spread plays · as of 2026-09-02T02:48:08.572892+00:00	model	20	—	0-0-20	—
[Auto] Soccer Spread (KNN) n=20 graded spread plays · as of 2026-09-02T02:48:08.572892+00:00	model	20	—	0-0-20	—
[Auto] Soccer Spread (MARS) n=20 graded spread plays · as of 2026-09-02T02:48:08.572892+00:00	model	20	—	0-0-20	—
[Auto] NHL Puck Line (Gaussian Process) n=7 graded spread plays · as of 2026-08-28T01:40:30.487538+00:00	model	7	—	0-0-7	—
[Auto] NHL Puck Line (HistGradientBoosting) n=7 graded spread plays · as of 2026-08-28T01:40:30.487538+00:00	model	7	—	0-0-7	—
[Auto] NBA Elo Power Ratings Spread (XGBoost) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (AdaBoost) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (Gaussian Process) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (HistGradientBoosting) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (KNN) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (MARS) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
[Auto] NBA Spread (SVR) n=5 graded spread plays · as of 2026-05-18T09:00:00.585548+00:00	model	5	—	0-0-5	—
Market consensus baseline · bet every favourite n=7,333 graded spread plays · as of 2026-10-18	baseline	7,333	48.7% · n=7,132 Backtest · all time · n=7,132 · through Oct 18, 2026	3473-3659-201	—

The baseline bets every favourite against the spread across past games. It has no closing-line value, so that cell stays empty.

Not ranked yet, no graded spread picks on record: community lens, pundit, crew.

Unlock more Insights tools

Keep exploring for free, or unlock the complete Insights toolkit with Pro.

$1 first week, then $2.99/week.

See Pro

Insights · real claim grader

The graded field: the market, mined

Every pundit claim in the transcript corpus, typed and graded against what actually happened. Who has been right, who has enough evidence to judge, and where the observed field sits against the house sample and fade gates.

25 decided grades across 2 real sources · all-time window · small samples are labeled as such. Coverage: top-decile and quadrant labels describe only the currently graded field, not ungraded or unresolved calls.

Claims graded as of Oct 17, 3:54 AM ET · Source registry as of Oct 13, 3:15 AM ET

Pundit claim grading

newest 6 decided claims · same rows as the ledger

Claim	Source	Verdict
“The Falcons somehow win primetime games this season”	Jeffrey · Oct 9	wrong
“but I'm still gonna pick the Ravens.”	Ben Solak (The Football Show) · Oct 29	wrong
“I'm gonna pick the Bills.”	Ben Solak (The Football Show) · Oct 29	correct
“I still think that the Rams have a better chance because of their consistency.”	Ben Solak (The Football Show) · Oct 29	wrong
“I still think the Jets would be able to win.”	Ben Solak (The Football Show) · Oct 29	correct
“I'm gonna pick the Browns.”	Ben Solak (The Football Show) · Oct 29	wrong

Pundit ledger

Decided, transcript-backed calls by real source. Pushes and unresolved claims never enter the hit-rate denominator.

correct and wrong calls only quoted from the transcript

Source	W-L	Hit %	n	Window	Verdict
Ben Solak (The Football Show)	14-10	58.3%	24	all-time	hot
early — not enough graded claims (1 sources)	0-1	0.0%	1	all-time	unproven

25 decided grades across 2 real sources · all-time window

Coverage: sources with n≤4 are collapsed into the early row; their grades remain in its W-L and n.

Verdict rules: unproven below n=20; gasbag below 45% after the gate; hot is the top decile of the gated field. With a thin gated field, hot is field-relative coverage, not a claim about ungraded calls.

Gasbag quadrant

More decided claims move right; a higher correct/wrong hit rate moves up. Every dot is a directly labeled real source.

correct + wrong only all-time

Coverage: 1 early sources are collapsed in the ledger and omitted here.

Counterfactual boundary: the guide lines classify observed grades only; they do not infer how a source would have performed on unresolved or unmade calls.