Advanced Data-Driven NBA Predictions
Subscribe to gain complete access.
Home of Estimated Plus-Minus (EPM)
EPM is a highly optimized, predictive player impact metric that leverages all available data (including player tracking data), incorporates knowledge about how each underlying stat stabilizes, and accounts for key predictive factors to estimate impact at any point in time. Here are the top 5 players right now. Click here to see more.
ESTIMATED +/- | ||||
---|---|---|---|---|
Nikola Jokic DEN ·
C
6'10" 250
·
29 | 37.5 99 | +7.4 100 | +1.2 89 | +8.5 100 |
Shai Gilgeous-Alexander OKC ·
G
6'6" 180
·
26 | 34.6 95 | +5.3 99 | +1.3 91 | +6.6 99 |
Luka Doncic DAL ·
G
6'6" 218
·
25 | 34.4 95 | +5.0 99 | +1.6 94 | +6.6 99 |
Joel Embiid PHI ·
C
7'0" 250
·
30 | 30.4 82 | +3.9 97 | +2.5 99 | +6.4 99 |
Jayson Tatum BOS ·
F
6'8" 205
·
26 | 36.3 98 | +4.8 99 | +1.0 86 | +5.8 99 |
Points per 100 Poss.
Machine-Learned Projection System
The stats that feed EPM are individually optimized via machine learning to weight past games appropriately in predicting next game values. These stats handle the question about sample size for you by distinguishing real change from random fluctuations. The result is an estimate of true skill at each point in time. Here is Anthony Edward's estimated skill in points / 100 possessions over his career.
Team EPM: Player-Driven Team Ratings
Team ratings powered by EPM offer greater accuracy than traditional team ratings, and are more flexible at the start of the season, when there are injuries, and following trades. Additionally, the Strength of Schedule is more precise, reflecting the players who participated in previous games. Here are the top 5 teams right now. Click here to see more.
Final | Win Prob. | pScore | Score | ||
---|---|---|---|---|---|
13 | Cavaliers | 24% | 101 | 98 | |
1 | Celtics | 76% | 112 | 113 |
Final | Win Prob. | pScore | Score | ||
---|---|---|---|---|---|
6 | Mavericks | 36% | 109 | 104 | |
2 | Thunder | 64% | 114 | 92 |
EPM-based Game Predictions
A game-optimized version of EPM is used to generate win probabilities for all regular season and playoff games. Game predictions update based on available information about who will play in the game and are more accurate than traditional ratings in predicting game outcomes. Click here to see current game predictions.
Boxscore Predictions
Optimized predictions for each player stat and game are made possible by the machine-learned estimated skill models which consider entire player careers, how game context influences each stat individually, and understand the difference between random noise and ability. Here is an abbreviated boxscore, with predicted values below actual values, and green circles indicating an exact hit.
Game | MIN | PTS | REB | AST | 3P | 3P% |
---|---|---|---|---|---|---|
Jaylen Brown G-F
·
6'7" 225
·
27.6 | 41 35 | 11 23 | 4 6 | 7 3 | 1-3 2-5 | 33% 35% |
Jayson Tatum F
·
6'8" 205
·
26.2 | 43 41 | 25 28 | 10 10 | 9 5 | 3-7 3-7 | 43% 37% |
Al Horford F-C
·
6'10" 245
·
37.9 | 35 30 | 22 9 | 15 7 | 5 2 | 6-13 2-5 | 46% 36% |
Derrick White G
·
6'4" 190
·
29.9 | 36 32 | 18 14 | 5 4 | 6 4 | 4-9 3-7 | 44% 37% |
Jrue Holiday G
·
6'4" 205
·
33.9 | 41 38 | 13 11 | 3 6 | 4 5 | 1-4 2-5 | 25% 36% |
Additional Features
Playoff Probabilities
Predicted win totals, playoff seeding probabilities, and the chance of winning each round is produced from 100,000 nightly simulations using EPM-powered game prediction models.
Powerful Interactive Dashboards
Intuitively command your view of the data by clicking on dynamic tables and beautiful visualizations for quick insight into player, team, and game predictions for any stat at any point in time.
Dynamic Player Comparison Tool
Compare EPM, estimated skills, and traditional stats for up to 5 players at a time by age, season, or career game number. Compare across player careers and at any point in time with percentiles and data visualizations.
Leverage Advanced NBA Predictions and Powerful Interactive Visuals
Subscribe to gain complete access.