Expected score models expose tactical reality behind losing streaks

Kenny Atkinson claimed his Cavaliers had 'analytically' won two of the first three games in the 2026 Eastern Conference Finals despite trailing 3-0, but a reconstruction of expected scores using actual shot attempts valued at season-long shooting rates showed the Knicks held an expected margin of 17 points across the series (Knicks News). The Knicks actually outscored Cleveland by 77 points, as the Cavaliers shot 47 points below their regular-season norms while the Knicks performed consistently with their season averages.

This reconstruction method takes every two-point, three-point, and free-throw attempt from a game and assigns the value of each based on a team's established 2025-26 shooting rates, such as Cleveland's .578 on twos and .360 on threes (Knicks News). Because this model treats every attempt as an average shot without accounting for defensive pressure, it provides a generous baseline for the offense, yet the Cavaliers still lost the expected score in the four-game series by 17 points.

Teams often cite favorable expected margins to explain away losses, but applying this standardized valuation strips away the noise of high-variance events to reveal systematic underperformance. The Knicks maintained their regular-season offensive efficiency of 119.2 points per 100 possessions against Cleveland, while the Cavaliers dropped from a 119.2 season average to 99.9 during the series.

• The Texans increased their 11-personnel usage to 89.1% in a single game to leverage their wide receiver depth, yielding a +0.10 EPA/play efficiency (Sam Hoppen from Hoppen to Conclusions).
• Devils defenseman Alex Huang maintains a 65% on-ice shot share at 5v5 by transitioning between recovering, closing, retrieving, and attacking states during defensive shifts (Jack Han from Hockey Tactics Newsletter).