Edge Lab · choose a source-backed question
Run a saved question, or type free text naming home dogs, division games, primetime, wind, rest-day edge, week, public tickets, line move, spread, or total. If the graded history can't score what you asked, you get "Couldn't match this question", never a different question's number.
Every answer is a count of what happened against the closing line, not a priced bet. A ?query= link reopens the same saved answer and is marked stale once it is more than
a day old. Selecting or running a question does not award XP.
Edge · Lab · Bench
Leaderboard
Every model with a live graded spread record stays on the board. Models with fewer than 30 decided games stay visible below the ranked field.
Ranked models
Ranked by ATS record, then closing-line value, then decided sample size.
Minimum sample: 30 decided games
| Rank | Model | Record | CLV | Training calibration gap | n | Window | As of |
|---|---|---|---|---|---|---|---|
| #1 | NFL Spread (Adaptive spline regression) | 32-18-0 64.0% | -0.02 pts | — | 50 | All live pregame grades | Oct 18, 2026 |
| #2 | NFL Spread (Adaptive spline regression) | 36-24-21 60.0% | +0.45 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #3 | Momentum Spread Model (Elastic net regression) | 39-29-13 57.4% | +0.07 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #4 | NFL Elo Power Ratings Spread (Gradient boosted trees) | 35-27-0 56.5% | +0.02 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #5 | NFL Spread (Support vector regression) | 34-27-20 55.7% | +0.39 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #6 | NFL Spread (Randomized trees) | 31-25-25 55.4% | +0.12 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #7 | NFL Elo Power Ratings Spread (Gradient boosted trees) | 36-30-15 54.5% | +0.12 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #8 | NFL Spread (Nearest neighbors) | 31-26-24 54.4% | +0.22 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #9 | Momentum Spread Model (Gradient boosted trees) | 34-29-18 54.0% | -0.02 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #10 | Full Feature Spread (Stacking Ensemble) | 30-26-25 53.6% | +0.48 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #11 | Turnover Regression Model (Elastic net regression) | 30-27-24 52.6% | +0.30 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #12 | Full Feature Spread (Stacking Ensemble) | 32-29-0 52.5% | +0.02 pts | — | 61 | All live pregame grades | Oct 18, 2026 |
| #13 | NFL ATS Gradient boosted trees (Custom Objective) | 33-30-18 52.4% | +0.30 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #14 | NFL Spread (HistGradientBoosting) | 30-28-23 51.7% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #15 | NFL Spread (Nearest neighbors) | 31-29-0 51.7% | -0.04 pts | — | 60 | All live pregame grades | Oct 18, 2026 |
| #16 | NFL GNN Matchup | 32-30-0 51.6% | +0.10 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #17 | NFL GNN Matchup | 29-28-24 50.9% | +0.48 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #18 | Full Feature Spread (Mixture of Experts) | 31-30-0 50.8% | +0.01 pts | — | 61 | All live pregame grades | Oct 18, 2026 |
| #19 | NFL Spread Stacking | 33-32-16 50.8% | +0.39 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #20 | Full Feature Spread (Gradient boosted trees) | 32-32-17 50.0% | +0.26 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #21 | Full Feature Spread (Mixture of Experts) | 31-31-0 50.0% | +0.09 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #22 | NFL Spread (Support vector regression) | 39-39-0 50.0% | +0.01 pts | — | 78 | All live pregame grades | Oct 18, 2026 |
| #23 | NFL ATS Gradient boosted trees (Custom Objective) | 31-31-0 50.0% | 0.00 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #24 | NFL Spread (AdaBoost) | 29-30-22 49.2% | +0.22 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #25 | Pressure & Efficiency Model (LightGBM) | 31-33-17 48.4% | +0.12 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #26 | Full Feature Spread (Mixture of Experts) | 27-29-25 48.2% | +0.24 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #27 | Feature Store Spread Model (Gradient boosted trees) | 22-24-0 47.8% | +0.02 pts | — | 46 | All live pregame grades | Oct 18, 2026 |
| #28 | Full Feature Spread (Gradient boosted trees) | 29-32-20 47.5% | +0.33 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #29 | NFL Minimal Signal Spread (Gradient boosted trees) | 29-32-20 47.5% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #30 | NFL MLP Spread | 28-31-22 47.5% | +0.33 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #31 | NFL MLP Spread | 29-33-0 46.8% | +0.01 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #32 | Momentum Spread Model (Elastic net regression) | 37-43-1 46.3% | +0.06 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #33 | NFL Market Efficiency Spread (Gradient boosted trees) | 27-32-22 45.8% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #34 | NFL Multi-Task Spread | 29-35-17 45.3% | +0.44 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #35 | NFL Spread (Gaussian Process) | 28-34-19 45.2% | +0.36 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #36 | NFL Spread (CatBoost) | 28-34-0 45.2% | -0.06 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #37 | Turnover Regression Model (Elastic net regression) | 36-44-1 45.0% | +0.30 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #38 | Turnover Regression Model (Elastic net regression) | 36-44-1 45.0% | +0.25 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #39 | Pressure & Efficiency Model (LightGBM) | 36-44-1 45.0% | +0.16 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #40 | Full Feature Spread (Stacking Ensemble) | 26-32-23 44.8% | +0.39 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #41 | SHAP Explainable Elo Model (Linear) | 29-36-16 44.6% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #42 | Momentum Spread Model (Elastic net regression) | 27-35-0 43.5% | +0.04 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #43 | NFL Spread (Randomized trees) | 27-35-0 43.5% | 0.00 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #44 | Simple Elo Spread Model (Linear) | 29-39-13 42.6% | +0.10 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #45 | NFL Multi-Task Spread | 26-36-0 41.9% | +0.06 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #46 | NFL Spread (HistGradientBoosting) | 26-36-0 41.9% | +0.02 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #47 | Simple Elo Spread Model (Bayesian) | 24-35-22 40.7% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #48 | Pressure & Efficiency Model (LightGBM) | 25-37-0 40.3% | -0.06 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #49 | SHAP Explainable Elo Model (Linear) | 32-48-1 40.0% | +0.14 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #50 | Simple Elo Spread Model (Bayesian) | 23-35-23 39.7% | +0.18 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #51 | Simple Elo Spread Model (Linear) | 31-49-1 38.8% | +0.10 pts | — | 81 | All live pregame grades | Oct 18, 2026 |
| #52 | Turnover Regression Model (Elastic net regression) | 24-38-0 38.7% | +0.04 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #53 | Pressure & Efficiency Model (LightGBM) | 24-38-0 38.7% | -0.07 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #54 | Simple Elo Spread Model (Bayesian) | 23-39-0 37.1% | -0.10 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #55 | SHAP Explainable Elo Model (Linear) | 23-39-0 37.1% | -0.10 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #56 | Sharkai Starter: NFL Spread | 22-40-0 35.5% | -0.02 pts | — | 62 | All live pregame grades | Oct 18, 2026 |
| #57 | NFL Market Efficiency Spread (Gradient boosted trees) | 16-30-0 34.8% | -0.15 pts | — | 46 | All live pregame grades | Oct 18, 2026 |
| #58 | NFL Minimal Signal Spread (Gradient boosted trees) | 15-31-0 32.6% | -0.03 pts | — | 46 | All live pregame grades | Oct 18, 2026 |
Insufficient sample
Fewer than 30 live decided games. These models stay visible but are not assigned a rank.
No model is currently below the 30-decided-game threshold.
Quests · claims and evaluations
No data
No claim quest progress yet
Sign in to track claim/evaluation quests. Missing progress is never converted to XP or a fabricated count.
- NO RUN No run has produced a measurement yet, so no guardrail can speak.
Coach
Ask a question → the Lab counts how it did over every graded game in the window. Read the caption → games counted, seasons covered, and the as-of date sit under every number. Compare on the Bench → calibration and leaderboard count their own games, so their numbers are not this one.