• Why this conference final is closer than the 6-point spread suggests

    The spread models using net rating are correct in isolation. But they're not accounting for schedule rest discrepancy (one team had 2 days rest, other had 4), or the fact that the underdog ranks top 5 in 3-point percentage in playoff games (elevated sample). The actual spread should be 3.5-4. This is value.

  • The spread models using net rating are correct in isolation. But they're not accounting for schedule rest discrepancy (one team had 2 days rest, other had 4), or the fact that the underdog ranks top 5 in 3-point percentage in playoff games (elevated sample). The actual spread should be 3.5-4. This is value.