NBA Raptor Rating Calculator
Estimate a player’s Raptor Rating based on key performance metrics and contextual factors
Raptor Rating Results
Estimated Raptor Rating:
—
Offensive Raptor:
—
Defensive Raptor:
—
Estimated League Rank:
—
Performance Tier:
—
How Are Raptor Ratings Calculated in the NBA?
The Raptor Rating (or simply “Raptor”) is an advanced basketball metric developed by FiveThirtyEight that measures a player’s total on-court impact in points above league average per 100 possessions. Unlike traditional box score statistics, Raptor incorporates tracking data, play-by-play information, and complex regression models to provide a comprehensive view of player performance.
Understanding the Raptor Rating System
Raptor was introduced in 2019 as an evolution of FiveThirtyEight’s earlier CARMELO projection system. The metric is designed to:
- Measure both offensive and defensive contributions separately
- Account for the quality of teammates and opponents
- Adjust for game situations (score margin, time remaining)
- Incorporate luck-adjusted shooting percentages
- Provide context-neutral evaluations of player performance
Key Components of Raptor
Raptor consists of several interconnected components that work together to produce the final rating:
- Box Score Prior: Uses traditional statistics (points, rebounds, assists, etc.) to establish a baseline expectation for each player’s performance.
- Tracking Data: Incorporates SportVU/Second Spectrum data including player movement, shot locations, defensive positioning, and more.
- Play-by-Play Adjustments: Considers the sequence of events in each possession to properly attribute credit/blame.
- Teammate/Opponent Adjustments: Accounts for the quality of players shared the court with and against.
- Luck Adjustments: Filters out random variation in shooting percentages and other outcomes.
- Positional Adjustments: Normalizes ratings across different positions with different typical responsibilities.
The Mathematical Foundation of Raptor
At its core, Raptor uses a regularized regression framework (similar to Ridge Regression) to estimate each player’s impact while preventing overfitting to noisy data. The model can be expressed conceptually as:
Raptor = Box Score Prior + Tracking Data Contributions + Contextual Adjustments + Random Effects
The box score prior serves as a Bayesian anchor that prevents the model from giving extreme ratings to players with limited playing time or unusual statistical profiles. This prior is particularly important for:
- Rookie players with no NBA track record
- Players returning from injury with limited recent data
- Role players with specialized but limited skill sets
- Players who change teams/roles significantly
Offensive vs. Defensive Raptor
Raptor separates offensive and defensive contributions, which are then combined for the total rating:
| Component | Description | Key Inputs | Typical Range (per 100) |
|---|---|---|---|
| Offensive Raptor | Measures a player’s contribution to their team’s offensive efficiency | Scoring volume, efficiency, playmaking, offensive rebounding, foul drawing | -5 to +15 |
| Defensive Raptor | Measures a player’s contribution to their team’s defensive efficiency | Rim protection, perimeter defense, defensive rebounding, foul avoidance | -10 to +5 |
| Total Raptor | Combined offensive and defensive impact | Sum of offensive and defensive components | -15 to +20 |
| WAR (Wins Above Replacement) | Converts Raptor into estimated wins contributed | Raptor rating + minutes played + league pace | 0 to 20+ |
Notably, defensive Raptor tends to have a narrower range than offensive Raptor because:
- Defensive impact is generally harder to measure with available data
- Most defensive contributions are team-dependent (scheme matters more)
- There’s less variation in defensive performance across players
- The model is more conservative with defensive estimates to avoid overstating impact
How Raptor Differs From Other Advanced Metrics
While Raptor shares some conceptual similarities with other advanced metrics, it has several distinctive features:
| Metric | Data Sources | Strengths | Weaknesses | Raptor Comparison |
|---|---|---|---|---|
| PER (Player Efficiency Rating) | Box score only | Simple, widely available, position-adjusted | Overvalues scoring volume, ignores defense | Raptor uses PER-like box score prior but adds much more |
| WS (Win Shares) | Box score + team performance | Intuitive “wins” framework, accounts for team success | Team-dependent, credit allocation issues | Raptor avoids team-success contamination through adjustments |
| BPM (Box Plus/Minus) | Box score + team +/- | Better than PER at valuing non-scoring contributions | Still box-score dependent, team +/- noise | Raptor replaces team +/- with tracking data |
| EPM (Estimated Plus/Minus) | Box score + tracking data | Incorporates some tracking data, publicly available | Less sophisticated adjustments than Raptor | Raptor has more complex modeling and adjustments |
| LEBRON (Luck-adjusted ERating) | Box score + luck adjustments | Accounts for shooting luck, simple to understand | No tracking data, limited defensive evaluation | Raptor includes LEBRON-like luck adjustments plus more |
Why Tracking Data Matters
The incorporation of SportVU/Second Spectrum tracking data is what truly sets Raptor apart from earlier metrics. This data includes:
- Player movement: Speed, distance traveled, changes of direction
- Defensive positioning: Distance from shooter, closeout speed, contest quality
- Shot location data: Exact (x,y) coordinates of every shot attempt
- Passing networks: Who passes to whom, where passes originate/land
- Screening data: Screen location, effectiveness, defender reaction
- Rebounding positioning: Where players are when shots go up
For example, tracking data allows Raptor to:
- Distinguish between a contested three-pointer and a wide-open one
- Measure how often a defender forces drivers into help defense
- Quantify the value of off-ball movement that doesn’t show up in box scores
- Evaluate screening effectiveness beyond just “screen assists”
- Assess defensive closeout quality on jump shooters
The Raptor Calculation Process Step-by-Step
While the exact proprietary formula isn’t public, we can outline the general calculation process based on FiveThirtyEight’s methodology descriptions:
- Data Collection: Gather box score stats, tracking data, and play-by-play information for every possession in the NBA season.
- Possession Segmentation: Break down each possession into discrete events (passes, drives, shots, etc.) with responsible players.
- Expected Value Calculation: For each event, calculate the expected point value based on historical data (e.g., expected points for a corner 3 vs. a mid-range jumper).
- Player Attribution: Allocate credit/blame for the difference between expected and actual outcomes to involved players using the regression model.
- Contextual Adjustments: Adjust for:
- Teammate quality (via “lineup priors”)
- Opponent quality (via opponent adjustments)
- Game situation (score, time remaining)
- Positional roles (different expectations for PG vs. C)
- Luck Adjustment: Filter out random variation in shooting percentages and other high-variance outcomes.
- Regression Modeling: Run the regularized regression to produce stable estimates for each player’s offensive and defensive impact.
- Scaling: Convert the raw impacts into points per 100 possessions above/below league average.
- WAR Conversion: (Optional) Convert the per-possession ratings into total wins above replacement based on minutes played.
Example Calculation
Let’s walk through a simplified example for a single possession:
Scenario: Player A drives to the basket, draws help defense, and kicks out to Player B for an open corner three that Player B makes.
- Initial State: The possession starts with an expected value of ~1.1 points (league average).
- Drive Creation: Player A’s drive increases the expected value to 1.15 points (creating a slight advantage).
- Pass to Corner: The kickout to the corner increases expected value to 1.25 points (corner threes are high-value shots).
- Made Shot: The made three is worth 3 points, so there’s a +1.75 point surplus over initial expectation.
- Credit Allocation: The model might allocate:
- +0.8 to Player A (for creating the advantage)
- +0.9 to Player B (for making the shot)
- +0.05 to other teammates (for spacing/off-ball movement)
- Defensive Debits: The defenders would receive corresponding negative credit for allowing this outcome.
This process repeats for every possession, with the model learning which player actions consistently lead to positive/negative outcomes.
Strengths and Limitations of Raptor
Like any advanced metric, Raptor has both significant advantages and important limitations that users should understand.
Strengths of Raptor
- Comprehensive Data Usage: Incorporates more data sources than any publicly available metric, including tracking data that captures aspects of the game box scores miss.
- Contextual Adjustments: Accounts for teammate quality, opponent quality, and game situations better than most alternatives.
- Defensive Evaluation: Provides more nuanced defensive ratings than box-score-based metrics.
- Luck Adjustments: Filters out random variation in shooting percentages that can distort other metrics.
- Positional Normalization: Allows for fairer comparisons across positions with different typical roles.
- Public Transparency: While the exact formula isn’t public, FiveThirtyEight has been relatively transparent about the methodology compared to some proprietary metrics.
- Historical Comparisons: The model can place current players in historical context by accounting for era differences in pace, efficiency, etc.
Limitations of Raptor
- Black Box Nature: The exact formula and weightings aren’t public, making it difficult to audit or fully understand the calculations.
- Tracking Data Limitations: While better than box scores, current tracking data still misses some important aspects of basketball (e.g., detailed off-ball defense, some types of screens).
- Small Sample Issues: Like all advanced metrics, Raptor can be noisy for players with limited minutes or in unusual roles.
- Defensive Challenges: Defense remains harder to measure than offense, and Raptor’s defensive ratings should be treated with appropriate caution.
- Scheme Dependence: Players in certain systems (e.g., extreme pace-and-space offenses) may be rated differently than they would be in other contexts.
- Injury Adjustments: The model may not fully account for players returning from injury or playing through injuries.
- Clutch Performance: While game situation is considered, Raptor may not fully capture “clutch gene” effects that manifest in high-leverage moments.
How Teams Use Raptor in Decision Making
NBA front offices have increasingly incorporated Raptor and similar advanced metrics into their decision-making processes. Some key applications include:
- Draft Evaluation: Teams use Raptor projections to identify underrated prospects whose box score stats don’t tell the full story (e.g., defensive specialists or high-IQ players who don’t score much).
- Free Agency Targeting: Front offices look for players whose Raptor ratings suggest they’re underpaid relative to their impact, or who might flourish in a different system.
- Trade Evaluation: Raptor helps quantify the value difference between players in potential trades, accounting for both current performance and projected aging curves.
- Lineup Optimization: Coaches and analysts use Raptor ratings to identify which player combinations work best together and against which opponents.
- Contract Negotiations: Agents and teams use Raptor (among other metrics) to argue for certain contract values, especially for role players whose impact isn’t obvious from box scores.
- Development Focus: Player development staff use the component breakdowns to identify specific skills players should work on to maximize their impact.
- Opponent Scouting: Teams analyze opponents’ Raptor ratings to identify weaknesses to exploit and strengths to neutralize.
For example, a team might use Raptor to:
- Identify that a player with modest counting stats actually has elite defensive impact that’s being undervalued
- Recognize that a high-scoring player is actually hurting their team’s offense due to inefficient shot selection
- Discover that certain five-man lineups are dramatically more effective than others
- Project how a player’s performance might change if they were traded to a team with a different system
Case Study: The 2019 Toronto Raptor Championship
The 2018-19 Toronto Raptors provide an interesting case study in how Raptor ratings can identify championship-caliber teams. That season:
- Kawhi Leonard led the NBA with a +8.5 Raptor rating (2nd in offensive Raptor, 12th in defensive Raptor)
- Pascal Siakam had a breakout season with +4.3 Raptor (showing his two-way impact)
- The Raptors had 5 players with positive Raptor ratings in their playoff rotation
- Their bench (led by Fred VanVleet at +3.1 Raptor) was one of the deepest in the league
- The team’s overall Raptor-based projected win total was 58.7 (they won 58 games)
Notably, Raptor identified several key advantages for Toronto:
- Their ability to switch defensively without major drop-offs
- Kawhi’s elite two-way impact that wasn’t fully captured by traditional stats
- The underrated defensive contributions of players like Marc Gasol and Danny Green
- The efficient offense generated by their balanced attack
This aligns with their eventual championship run, where these exact factors proved decisive in the playoffs.
Common Misconceptions About Raptor
Despite its sophistication, there are several common misunderstandings about what Raptor does and doesn’t measure:
- “Raptor is just a fancy plus/minus”: While it uses some similar concepts, Raptor incorporates far more data and adjustments than traditional plus/minus metrics.
- “It ignores clutch performance”: Raptor does account for game situations, though like all metrics it may not fully capture the psychological aspects of “clutchness.”
- “It’s only about offense”: Raptor actually provides separate offensive and defensive ratings, with the defensive component being particularly sophisticated.
- “It can’t measure leadership”: True – Raptor focuses on on-court impact. Leadership and other intangibles require separate evaluation.
- “Higher Raptor always means better player”: Context matters – a +5 Raptor center and a +5 Raptor point guard contribute differently to team success.
- “It’s perfectly accurate”: All metrics have error bars. Raptor is better than most but still an estimate with uncertainty.
- “It replaces scouting”: Smart teams use Raptor alongside traditional scouting, not as a replacement.
How to Interpret Raptor Ratings
Understanding how to properly interpret Raptor ratings is crucial for getting value from the metric:
Reading the Numbers
Raptor ratings are expressed as points per 100 possessions above or below league average:
- +10 or higher: MVP-caliber season
- +6 to +10: All-NBA level
- +3 to +6: All-Star level
- +1 to +3: Solid starter
- -1 to +1: Rotation player
- -3 to -1: Bench contributor
- Below -3: Replacement level or worse
For context, here are some recent single-season Raptor leaders:
| Season | Player | Total Raptor | Offensive Raptor | Defensive Raptor | WAR |
|---|---|---|---|---|---|
| 2022-23 | Nikola Jokić | +12.1 | +10.8 | -0.3 | 15.3 |
| 2021-22 | Nikola Jokić | +11.8 | +11.2 | +0.6 | 15.7 |
| 2020-21 | Joel Embiid | +10.7 | +8.9 | +1.8 | 12.9 |
| 2019-20 | LeBron James | +9.8 | +8.5 | +1.3 | 12.1 |
| 2018-19 | James Harden | +9.6 | +11.1 | -1.5 | 13.6 |
| 2017-18 | LeBron James | +9.3 | +7.8 | +1.5 | 12.4 |
Contextual Factors to Consider
When evaluating Raptor ratings, always consider:
- Minutes Played: Ratings for players with <1000 minutes in a season have wider error bars.
- Role Changes: A player’s rating may change significantly with a new role (e.g., moving from bench to starter).
- Team System: Players in certain systems (e.g., D’Antoni’s offense) may have inflated/deflated ratings.
- Injury Status: Players returning from injury may have depressed ratings early in their return.
- Age: Younger players often improve, while older players typically decline (Raptor accounts for this but isn’t perfect).
- Position: A +5 rating means different things for a point guard vs. a center.
- Era: League-wide efficiency changes over time affect what constitutes an “elite” rating.
Combining Raptor with Other Metrics
For the most complete player evaluation, Raptor should be considered alongside:
- Traditional Stats: Points, rebounds, assists provide context for how players accumulate their impact.
- Shooting Splits: True shooting percentage, free throw rate, and shot location data help explain offensive ratings.
- Play Type Data: Synergy or NBA.com play type stats show how players generate their offense.
- On/Off Data: Team performance with/without the player on court can corroborate Raptor findings.
- Eye Test: Watching games helps identify intangibles and contextual factors metrics might miss.
- Other Advanced Metrics: BPM, VORP, and LEBRON can provide alternative perspectives.
The Future of Raptor and Advanced Metrics
As basketball analytics continue to evolve, we can expect several developments in metrics like Raptor:
- More Granular Tracking Data: New camera systems and wearables will provide even more detailed information about player movements, fatigue levels, and biometric responses.
- Improved Defensive Metrics: Future versions will likely better capture defensive positioning, communication, and scheme execution.
- Real-Time Analysis: Metrics may eventually be calculated and available in real-time during games.
- Better Injury Modeling: Incorporating medical data (where ethical) could improve projections for injured players.
- Psychological Factors: Future metrics might attempt to quantify “clutchness,” leadership, and other intangibles.
- Cross-League Comparisons: As international leagues get more tracking data, we may see global player comparisons.
- AI and Machine Learning: More sophisticated models may emerge that can detect patterns humans miss.
However, some fundamental challenges will remain:
- Basketball is a complex, fluid game that resists complete quantification
- Some aspects of player impact may always require human judgment
- The trade-off between model complexity and interpretability will persist
- Ethical considerations around data collection and usage will grow more important
How Fans Can Use Raptor
While NBA teams have access to more detailed versions, fans can use the public Raptor ratings to:
- Evaluate Trades: Compare the Raptor ratings of players involved in proposed trades to see which team might be getting better value.
- Assess Draft Prospects: Look at college/international players’ projected Raptor ratings to identify potential sleepers.
- Understand Lineup Decisions: See why certain players might be getting more/less playing time based on their impact metrics.
- Predict Awards: Raptor leaders often correlate with MVP, DPOY, and All-NBA selections.
- Identify Breakout Candidates: Look for players whose Raptor ratings are improving faster than their traditional stats.
- Evaluate Coaching Impact: Compare players’ Raptor ratings before/after coaching changes to assess scheme effects.
- Engage in Debates: Use Raptor as evidence in discussions about player value (while acknowledging its limitations).