Case Study 1: The Discovery of Draymond Green's True Value

Introduction

Few players in NBA history have better illustrated the gap between traditional statistics and true impact than Draymond Green of the Golden State Warriors. This case study examines how RAPM and related plus-minus methodologies revealed Green's elite value years before mainstream recognition, demonstrating the power of regression-based impact metrics.

Background: The Traditional Statistics Problem

Green's Career Arc

Draymond Green was selected 35th overall in the 2012 NBA Draft by the Golden State Warriors. His college statistics at Michigan State were solid but unspectacular: 10.6 PPG, 7.6 RPG, and 3.0 APG as a senior. These numbers suggested a useful role player, not a future All-Star or All-NBA selection.

In his first two NBA seasons (2012-13, 2013-14), Green's traditional statistics remained modest:

These numbers placed Green among the bottom half of NBA starters by scoring and shooting efficiency. Traditional analysis suggested a limited ceiling.

The Hidden Impact

What traditional statistics missed was everything else Green did on the basketball court:

• Defensive versatility: Ability to guard all five positions
• Help defense and rotations: Reading passing lanes, rim protection despite size
• Offensive facilitation: Creating opportunities for teammates through screens and passing
• Floor spacing effect: Commanding defensive attention without scoring
• Communication and leadership: Organizing defensive schemes

None of these contributions appeared in standard box scores.

RAPM Analysis: Revealing True Value

Early RAPM Signals

When analysts began calculating RAPM for the 2013-14 season, Green's numbers stood out dramatically:

Despite playing fewer minutes than other starters, Green's RAPM ranked among the team's best. His D-RAPM of +2.7 was particularly striking given his 6'6" height—traditionally too small for elite interior defense.

The Methodological Advantage

Why did RAPM capture Green's value when traditional stats missed it?

1. Lineup-Based Measurement

RAPM measures how the team performs with Green on the court versus off it, controlling for who else is playing. This captured: - Defensive schemes only possible with Green's versatility - Offensive flow improvements from his playmaking - The cumulative effect of his intangible contributions

2. Adjustment for Context

Green played significant minutes alongside Stephen Curry. Traditional stats might attribute the team's success to Curry's obvious production. RAPM's regression framework assigned credit based on marginal contribution, revealing Green's independent value.

3. Defensive Credit

Traditional defensive statistics (steals, blocks) capture only the most visible defensive plays. RAPM credited Green for: - Shots not taken due to his positioning - Passes not made because of his help defense - Points saved by defensive communication

Statistical Validation

The 2014-15 season provided natural validation. When David Lee was injured and Green became a full-time starter, the Warriors improved dramatically:

The Warriors went 67-15, the best record in franchise history. Green's increased minutes allowed RAPM to estimate his impact with greater precision:

2014-15 RAPM: - Total: +6.8 (top 15 in NBA) - O-RAPM: +2.1 - D-RAPM: +4.7 (top 5 in NBA)

Deep Dive: The Death Lineup

Composition and Context

The Warriors' "Death Lineup" featured: - Stephen Curry (PG) - Klay Thompson (SG) - Andre Iguodala (SF) - Harrison Barnes (PF) - Draymond Green (C)

This lineup, with Green playing center at 6'6", defied conventional basketball wisdom. Traditional analysis would predict disaster: no rim protection, easy post-up opportunities for opponents, and rebounding problems.

RAPM Evaluation

Instead, the Death Lineup posted historically dominant numbers:

RAPM analysis showed that Green's coefficient was highest when sharing the court with small-ball lineups:

Green's RAPM by Lineup Type: - Traditional (with standard center): +4.2 - Small ball (Green at center): +8.1 - Difference: +3.9 points per 100 possessions

Understanding the Mechanism

RAPM identified that small-ball unlocked additional Green value through:

1. Defensive Switching

With Green at center, the Warriors could switch every screen without mismatches. Green's unique ability to guard 1-5 made this possible, eliminating opponents' pick-and-roll advantages.

2. Pace and Space

Green's ball-handling and passing from the center position created offensive opportunities unavailable with traditional bigs. The offense generated more open shots and fast-break opportunities.

3. Rebounding Redistribution

While Green grabbed fewer rebounds individually, the lineup's superior defense meant fewer opponent field goal attempts, reducing the rebounding burden.

Broader Implications

The Emergence of Analytics-Friendly Roster Construction

Green's RAPM profile influenced league-wide roster construction:

Before Green (Traditional Model): - Centers valued for height, post scoring, rebounding - Defense measured by blocks and steals - "Small ball" considered gimmicky or emergency-only

After Green (Analytics Model): - Centers valued for switching ability, passing, versatility - Defense measured by overall impact (RAPM, DRTG) - Small-ball lineups deployed strategically for advantages

Contract and Award Implications

Green's RAPM profile influenced major decisions:

2015 Contract Negotiation: - Traditional stats suggested ~$10M annually - RAPM suggested elite value (~$18-20M) - Green signed 5-year, $82M extension - Market value exceeded contract; Warriors benefited

Award Recognition: - 2017 Defensive Player of the Year - Multiple All-NBA and All-Defensive Team selections - Recognition lagged RAPM indicators by ~2 years

Technical Analysis: Collinearity Challenges

The Curry-Green Correlation

One challenge in estimating Green's RAPM was his high correlation with Stephen Curry. Both played starter minutes on elite teams, creating potential collinearity.

Lineup Overlap Analysis: - Curry minutes: 2,846 - Green minutes: 1,730 - Shared minutes: 1,520 (88% of Green's time)

This correlation could theoretically inflate uncertainty in individual estimates.

Multi-Year Resolution

Using three-year RAPM (2013-16) helped separate individual effects:

Observations with Both: 45,000+ possessions Observations with Curry only: 12,000+ possessions Observations with Green only: 8,000+ possessions

The varied lineup combinations provided enough separation to estimate individual coefficients with reasonable precision:

While confidence intervals overlap, both players' lower bounds exceed +2.5, confirming both as clearly positive contributors.

Lessons Learned

For Analysts

1. Box score metrics miss half the game: Traditional statistics focus on ball-dominant actions; RAPM captures everything affecting winning.

2. Defense requires regression-based measurement: Without tracking who allowed baskets (not just who blocked shots), individual defense is essentially invisible.

3. Context matters: Players may have higher value in specific configurations; RAPM can identify these synergies.

4. Validate with natural experiments: Injuries and lineup changes provide natural tests of RAPM estimates.

For Teams

1. Trust the process: RAPM identified Green's value before it was obvious; teams that trusted analytics gained advantages.

2. Design systems around unique profiles: Once RAPM revealed Green's switching value, the Warriors designed their entire defense around it.

3. Contract leverage: Players with RAPM exceeding traditional value perception may be undervalued in the market.

For Fans and Media

1. Look beyond scoring: The best players may not be the highest scorers.

2. Question conventional wisdom: "You can't play a 6'6" center" was wrong.

3. Appreciate defensive contributions: Defense wins championships, but it's invisible without proper metrics.

Conclusion

The Draymond Green case study demonstrates RAPM's power to identify elite value hidden by traditional statistics. Green's defensive versatility, playmaking, and intangible contributions—invisible to box scores—created winning basketball. RAPM's regression framework, by measuring actual point differential impact while controlling for teammates and opponents, revealed this value years before mainstream recognition.

This case also illustrates RAPM's limitations: it identified that Green was valuable but couldn't initially explain why. The synthesis of RAPM with tracking data and qualitative analysis eventually provided the full picture. Modern player evaluation requires integrating multiple information sources, with RAPM serving as the foundational measurement of true impact.

Discussion Questions

1. Why did traditional defensive metrics (blocks, steals, defensive rebounds) fail to capture Green's defensive value?

2. How might the Warriors' front office have used RAPM during Green's 2015 contract negotiations?

3. What other players might currently be undervalued by traditional metrics but overvalue by RAPM?

4. How does collinearity between Curry and Green affect our confidence in individual RAPM estimates?

5. What changes in league play style can be traced to the analytical validation of Green's approach?

Data Analysis Exercise

Using publicly available RAPM data:

1. Calculate the difference between Draymond Green's RAPM and traditional metrics (PER, WS/48) rankings for seasons 2014-15 through 2018-19.

2. Identify other players with similar RAPM-vs-traditional discrepancies.

3. Analyze whether these players share common characteristics (position, playing style, team quality).

4. Predict which current players might be "the next Draymond Green" based on RAPM vs. traditional metric gaps.