๐Ÿปโฌ‡๏ธ๐Ÿ€

2019-2020 NCAAM Model Performance

Scope

All scored games in the selected league and season. AP Poll is excluded here.

Every ranked model predicted the same 4300 games (2019-11-28 to 2020-03-11, 15 weekly windows) out of sample: each week's games were predicted by models fit only on the season's earlier games.

Does recruiting help team predictions?
Men: yes, mostly early. Adding freshman recruiting rankings to the preseason estimate makes Elo's game predictions better by 6.6 milli-nats per game in weeks 1-4 (95% CI 4.2 to 9.1) and by 3.1 over the full season (2.3 to 3.9), on six seasons it never saw; in 2025-26, also unseen, by 10.0 and 3.7. Picks right in weeks 1-4: 76.3% with recruiting, 76.0% without, 74.6% carrying last season over.
Milli-nats: 0.001 of log loss per game, lower is better. Recruiting model report · Preseason prior backtest

Model Catalog · Player models · Preseason prior backtest (team Elo with and without the player-based start)

Model Performance Leaderboard

Ranked by full-season AUC. The final week is the season's last 7 played days (261 games, 2020-03-05 to 2020-03-11), predicted from every earlier game: a small sample that shows the latest form.

# Model AUC Acc Brier LogLoss AUC final wk Acc final wk
1 Margin Recency Margin Recency Margin regression with exponential recency weights on newer games. More โ†’ 0.791 71.8% 0.191 0.563 0.744 67.0%
2 Margin Margin Linear team-strength model fit on point differential instead of binary wins. More โ†’ 0.791 71.8% 0.191 0.563 0.742 67.8%
3 Points Off/Def Recency Points Off/Def Recency Off/def points regression with exponential recency weights. More โ†’ 0.790 71.5% 0.191 0.565 0.743 67.0%
4 Points Off/Def Points Off/Def Raw points regression with separate offensive and defensive team parameters. More โ†’ 0.790 71.4% 0.191 0.565 0.741 67.0%
5 Adjusted Context Blend Adjusted Context Blend Chronologically trained component blend with rest and signed venue context. More โ†’ 0.782 70.3% 0.195 0.578 0.736 67.0%
6 Elo Elo Streaming paired-comparison rating with recency baked into sequential updates. More โ†’ 0.778 70.9% 0.192 0.564 0.728 67.8%
7 Recency Ensemble Recency Ensemble Equal-logit blend of Elo, recency BT, recency margin, log-adjusted pyth, and recency points off/def. More โ†’ 0.775 70.1% 0.194 0.568 0.732 68.6%
8 Core Ensemble Core Ensemble Equal-logit blend of Elo, recency BT, recency margin, log-adjusted pyth, and points off/def. More โ†’ 0.775 70.2% 0.194 0.568 0.732 68.2%
9 Adjusted Efficiency Adjusted Efficiency Opponent-adjusted efficiency model with separate offensive and defensive components. More โ†’ 0.763 69.6% 0.204 0.601 0.728 67.8%
10 Log Adjusted Log Adjusted Log-scale adjusted efficiency model that downweights blowout leverage. More โ†’ 0.763 69.7% 0.204 0.602 0.728 67.8%
11 Bradley-Terry Recency Bradley-Terry Recency Static Bradley-Terry with exponential recency weights on newer games. More โ†’ 0.748 68.1% 0.204 0.593 0.714 66.3%
12 Bradley-Terry Bradley-Terry Static logistic paired-comparison model with one team strength parameter. More โ†’ 0.748 68.2% 0.204 0.593 0.710 66.3%
13 Pythagorean Pythagorean Pythagorean win expectation from raw points scored and allowed. More โ†’ 0.747 68.6% 0.208 0.604 0.734 67.8%
14 Avg Margin Baseline Avg Margin Baseline Predict from simple average scoring margin in the training window. More โ†’ 0.744 68.3% 0.206 0.597 0.726 67.8%
15 Dynamic Bradley-Terry Dynamic Bradley-Terry Time-evolving paired-comparison model with latent team strength drift. More โ†’ 0.744 67.3% 0.206 0.596 0.716 67.0%
16 Home Team Baseline Home Team Baseline Always favor the home team with a fixed prior. More โ†’ 0.633 63.3% 0.233 0.660 0.657 65.9%

Methodology

Out-of-sample windows

  • The season is split into 7-day windows ending on its last played day
  • Each window is predicted by models refit on all earlier games of the season
  • The first window starts after 21 days of games
  • Every model is scored on the same games; a model that cannot predict most of them is listed as not compared

Elo / Bradley-Terry

  • Elo: the published league Elo (preseason priors and carryover where configured), frozen at the window start
  • Bradley-Terry: static logistic fit on the earlier games
  • Recency variants weight games with a 120-day half-life in NCAA leagues (keeping the non-conference games that link conferences) and a 30-day half-life in pro leagues

Pythagorean and margin models

  • Raw, adjusted and log-adjusted Pythagorean from scores; efficiency from box-score possessions
  • Margin: ridge regression on point margin with a learned home advantage
  • Ensembles average the log-odds of five families

Baselines

  • Home Team: the home team wins 60% (50% at neutral sites)
  • Avg Margin: the higher average margin so far wins
  • Models should beat these to add value