DataDunkNBA
Home / Metrics / WEV: Winning Expected Value
Metric · Teams

WEV: Winning Expected Value

A defense-weighted grade of how much winning a team's profile is worth. It has put 25 of the last 26 champions in its top ten.

WEV: a contender filterWinning Expected Value weights defense twice as heavily as offense. 25 of the last 26 champions finished in its top ten.
Transcript

WEV is 0.30 offense plus 0.60 defense plus 0.10 clutch, so defense counts double. 25 of the last 26 champions finished in WEV's top ten. The one miss was the 2001 Lakers, at No. 11. It finds the real contenders. It doesn't pick the champion.

PieceWeight
DEV (defense)0.60
OEV (offense)0.30
CEV (clutch)0.10

The formula

WEV = 0.30 × OEV + 0.60 × DEV + 0.10 × CEV, on a 0 to 25 scale. Defense gets twice the weight of offense, and when the weights were re-fit from scratch on seasons the model never saw, the defensive weight came back at about 0.58.

For new scoring we use the version without the clutch term, (0.30 × OEV + 0.60 × DEV) ÷ 0.90, because two of the three clutch inputs are only known after a title is won.

A filter, not a picker

Across the full panel, 25 of 26 champions finished in WEV's top ten and 22 of 26 in its top five. The one miss was the 2001 Lakers, at No. 11. That makes WEV the right tool for one question: who is a real contender. It is not built to pick the single champion, and it doesn't.

Champion record from the full 805-team-season panel. Held out: a spec frozen on 2000-2015 found 7 of 8 champions from 2016-2023 in its top five.

Read more: The four tests preview.

All metrics →

Essays on this from the newsletter

The full research library →