How does EdgeLine compare?
Check the evidence: same games, saved forecasts and all outcomes. Compare weekly verified picks, full results and the performance study.
EdgeLine vs sportsbook lines and nfelo
EdgeLine recorded lower average spread error than the saved sportsbook lines in this 14-game sample. EdgeLine: 10.39 points; sportsbook: 10.86 points — approximately 4.3% lower error. The sportsbook recorded lower total error, and nfelo opening forecasts recorded slightly lower spread error.
Fixed snapshot as of October 17, 2026. Games: 13 from Week 5 and one from Week 6. Lower average error in points (MAE) is better.
| Forecast | Spread games | Spread error | Total games | Total error |
|---|---|---|---|---|
| EdgeLine | 14 | 10.39 | 14 | 10.36 |
| Saved sportsbook line | 14 | 10.86 | 14 | 9.75 |
| nfelo opening forecast | 14 | 10.36 | — | — |
| nfelo closing forecast | 14 | 10.57 | — | — |
Exploratory comparison: Same completed games; forecast cutoffs differ. nfelo data was retrieved after these games; original publication timestamps and revision history are unverified. EdgeLine versions 4.6.0 and 5.0.0 are pooled here and separated in the JSON. This descriptive sample does not establish statistical superiority, betting profitability or future performance. No paired probability or ROI comparison is available.
nfelo source data · Game-level evidence, model versions and RMSE
NFL regression and XGBoost comparison: Weeks 1–5 and forward research →
Completed forecasts saved before kickoff. Model and market probability scores use identical games and two-sided saved prices. Missing evidence is excluded; no invented odds or competitor forecasts.
What is ready?
NFL paired completed games: 15. UFC completed forecasts: 22.
Learned UFC challenger: INSUFFICIENT_SAMPLE. Minimum 200 valid archived fights, with separate training, calibration and unseen test periods.
College evidence: Updated 2026-10-18T19:06:20.716Z. CFBD: connected; prior-season diagnostic enrichment. Verified starting QB feed: not connected.
No automatic promotion. A candidate needs later unseen results and paired market evaluation.
Probability accuracy against the market
Lower Brier score and log loss are better. Zero observations means insufficient evidence. Versions and markets are kept separate.
| Sport | Forecast | Market | Decisions | Brier | Log loss |
|---|---|---|---|---|---|
| NFL | 4.6.0 / 100K support | spread | 0 | — | — |
| NFL | Saved market (same games) | spread | 0 | — | — |
| NFL | 4.6.0 / 100K support | total | 0 | — | — |
| NFL | Saved market (same games) | total | 0 | — | — |
| NFL | 5.0.0 / 100K support | spread | 0 | — | — |
| NFL | Saved market (same games) | spread | 0 | — | — |
| NFL | 5.0.0 / 100K support | total | 0 | — | — |
| NFL | Saved market (same games) | total | 0 | — | — |
| UFC | ufc-ensemble-3.1 / final | Winner | 4 | 0.3801 | 0.9917 |
| UFC | Saved market (same fights) | Winner | 4 | 0.3868 | 0.9857 |
| UFC | ufc-ensemble-3.1 / before market | Winner | 4 | 0.3832 | 0.9979 |
| UFC | ufc-ensemble-3.0 / final | Winner | 6 | 0.2509 | 0.6950 |
| UFC | Saved market (same fights) | Winner | 6 | 0.2090 | 0.6063 |
| UFC | ufc-ensemble-3.0 / before market | Winner | 6 | 0.2483 | 0.6899 |
| College | college-challenger-v1-8a0388828bb9 / baseline | spread | 101 | 0.2487 | 0.6904 |
| College | Saved market (same games) | spread | 101 | 0.2515 | 0.6962 |
| College | college-challenger-v1-8a0388828bb9 / baseline | total | 104 | 0.2569 | 0.7071 |
| College | Saved market (same games) | total | 104 | 0.2506 | 0.6943 |
| College | college-challenger-v1-8a0388828bb9 / market | spread | 101 | 0.2515 | 0.6962 |
| College | Saved market (same games) | spread | 101 | 0.2515 | 0.6962 |
| College | college-challenger-v1-8a0388828bb9 / market | total | 104 | 0.2506 | 0.6943 |
| College | Saved market (same games) | total | 104 | 0.2506 | 0.6943 |
| College | college-challenger-v1-8a0388828bb9 / ridge | spread | 101 | 0.2542 | 0.7032 |
| College | Saved market (same games) | spread | 101 | 0.2515 | 0.6962 |
| College | college-challenger-v1-8a0388828bb9 / ridge | total | 104 | 0.2595 | 0.7128 |
| College | Saved market (same games) | total | 104 | 0.2506 | 0.6943 |
| College | college-challenger-v1-8a0388828bb9 / boosting | spread | 101 | 0.2517 | 0.7004 |
| College | Saved market (same games) | spread | 101 | 0.2515 | 0.6962 |
| College | college-challenger-v1-8a0388828bb9 / boosting | total | 104 | 0.2604 | 0.7148 |
| College | Saved market (same games) | total | 104 | 0.2506 | 0.6943 |
Scoring prediction errors
MAE is average error in points; RMSE gives larger misses more weight. These are not betting-return measures.
| Sport | Forecast | Market | Games | MAE | RMSE |
|---|---|---|---|---|---|
| NFL | 4.6.0 / current | spread | 13 | 11.0000 | 13.8439 |
| NFL | 4.6.0 / previous | spread | 13 | 10.6923 | 13.8578 |
| NFL | 4.6.0 / market | spread | 13 | 11.2308 | 13.9215 |
| NFL | 4.6.0 / current | total | 13 | 9.8462 | 10.4661 |
| NFL | 4.6.0 / previous | total | 13 | 10.1154 | 10.9166 |
| NFL | 4.6.0 / market | total | 13 | 9.7692 | 10.4274 |
| NFL | 5.0.0 / current | spread | 2 | 3.0000 | 3.0414 |
| NFL | 5.0.0 / previous | spread | 2 | 3.2500 | 3.2596 |
| NFL | 5.0.0 / market | spread | 2 | 6.0000 | 6.0000 |
| NFL | 5.0.0 / current | total | 2 | 16.0000 | 16.0312 |
| NFL | 5.0.0 / previous | total | 2 | 15.7500 | 15.8469 |
| NFL | 5.0.0 / market | total | 2 | 16.0000 | 17.2699 |
| College | college-challenger-v1-8a0388828bb9 / baseline | spread | 104 | 12.2750 | 14.9774 |
| College | college-challenger-v1-8a0388828bb9 / baseline | total | 104 | 12.7923 | 15.9413 |
| College | college-challenger-v1-8a0388828bb9 / market | spread | 104 | 11.1538 | 13.9482 |
| College | college-challenger-v1-8a0388828bb9 / market | total | 104 | 12.0865 | 14.9852 |
| College | college-challenger-v1-8a0388828bb9 / ridge | spread | 104 | 11.4901 | 14.5091 |
| College | college-challenger-v1-8a0388828bb9 / ridge | total | 104 | 12.5411 | 15.7218 |
| College | college-challenger-v1-8a0388828bb9 / boosting | spread | 104 | 11.6443 | 14.6694 |
| College | college-challenger-v1-8a0388828bb9 / boosting | total | 104 | 12.6079 | 15.6687 |
UFC probability calibration
All eligible completed versions; use the JSON report for version-specific results.
| Probability band | Fights | Average estimate | Actual win rate |
|---|---|---|---|
| 0.0%–10.0% | 0 | — | — |
| 10.0%–20.0% | 1 | 19.6% | 100.0% |
| 20.0%–30.0% | 0 | — | — |
| 30.0%–40.0% | 0 | — | — |
| 40.0%–50.0% | 7 | 45.9% | 28.6% |
| 50.0%–60.0% | 9 | 53.6% | 66.7% |
| 60.0%–70.0% | 3 | 62.3% | 33.3% |
| 70.0%–80.0% | 0 | — | — |
| 80.0%–90.0% | 2 | 86.6% | 50.0% |
| 90.0%–100.0% | 0 | — | — |
Research strategy: backing each forecast favorite at its saved offered price. Stakes: 10; units: -0.6560; ROI: -6.6%. This is separate from the official-play record. Missing prices are excluded.
Competitor comparisons
A fixed 14-game descriptive comparison with nfelo is published above. Prospective external rankings require forecasts archived before kickoff at comparable times.
Methodology research is not a verified accuracy ranking. Small samples and historical quote gaps remain visible.
Full metrics, calibration, returns and model versions (JSON) · NFL official-play validation · UFC validation · College challenger research
Updated 2026-10-18T19:08:16.809Z.