Who's the NBA's
best defensive player?
My model's pick for 2026-27 Defensive Player of the Year is Victor Wembanyama at 58.7%. Vegas has him around three in four, so the model thinks Vegas is too sure. It was tested on 42 seasons using only what was known on opening night.
Who protects the rim best right now? See the rim protection leaders.
Watch all five versions get built, the lab, and the countdown. Open on YouTube →
Victor Wembanyama. Vegas has him at −360, about three in four, and that's before the 65-game rule, which he only just met last season.
Seasons since 1985 where the model's preseason favourite won. The real winner was in its top three 22 times.
Favourites that won over the last 11 seasons, 67% more winners. But Vegas gave the real winner 15.8% on average vs the model's 13.2%: the model is better at picking, Vegas at pricing.
Even with every player's real full-season stats, something nobody knows in October, a model only picks half the winners. Voters still surprise it.
Every player's chance
The final model's preseason chance for the top 30 candidates, as shown in the video. Tap a column to sort. Names link to each player's stats page.
Update, Nov 1: Chet Holmgren is out 8–10 weeks and Cason Wallace is likely out most of the season, so neither can reach 65 games. Taking them off the board without refitting the model: Victor Wembanyama 63.6%, Ausar Thompson 2.8%, Shai Gilgeous-Alexander 2.5%. The preseason call above stays the official prediction.
Five versions
Each version is backtested: it goes back to every season since 1985, hides the answer and sees only what was known on opening night. Think of it as old exams with the answers covered up. The tag is how many of the 42 winners its favourite picked.
V1 · Last year's ballot 10 / 42
The laziest model possible: it assumes last year's vote happens again. It does well because this award repeats a lot, but it can never pick somebody new.
2026-27 top 3: Victor Wembanyama 55.6%, Chet Holmgren 26.6%, Ausar Thompson 6.7%
V2 · The best defender on paper 8 / 42
Projects every player's blocks, steals, defensive box plus-minus and defensive win shares from his last three seasons. It got worse: the best defender on paper isn't the guy voters pick.
2026-27 top 3: Victor Wembanyama 24.8%, Shai Gilgeous-Alexander 4.0%, Chet Holmgren 3.7%
V3 · What voters reward 11 / 42
Adds the things voters care about: an elite team defense, big minutes and enough games. 18 of the last 19 winners played on a top-three defense.
2026-27 top 3: Victor Wembanyama 58.6%, Shai Gilgeous-Alexander 8.7%, Chet Holmgren 7.5%
V4 · + Chaos 9 / 42
Simulates every season thousands of times: breakouts, injuries and team defenses that miss. More realistic odds, but it didn't pick more winners.
2026-27 top 3: Victor Wembanyama 33.4%, Shai Gilgeous-Alexander 9.9%, Chet Holmgren 8.7%
V5 · The lab (final) 14 / 42
Every idea had to improve the 42-season backtest or it was cut. The survivors are all about memory: last year's vote, finishing 2nd or 3rd last year, All-Defensive votes last year and two years ago, and “virtual memory” for players who missed a season injured. Then projected team defense, defensive box plus-minus, blocks, defensive rebounding and minutes.
2026-27 top 3: Victor Wembanyama 58.7%, Chet Holmgren 5.9%, Ausar Thompson 2.6%
What didn't make it
Most ideas died: a boost for the defending winner, a boost for big men, steals, defensive win shares, age, switching teams, even how good his team's defense was the year before. None of them picked more winners.
Virtual memory. In 2024-25 Wembanyama missed half the season with a blood clot, so he got zero votes and zero All-Defensive teams. Without a fix, the model ranked him 11th going into the year he won unanimously. Running his real 2024-25 numbers through the voter model says he'd have got about 40% of the vote if healthy.
Thrown out: a version that picked 16 of 42. I found it by trying every combination on the same 42 seasons, which is like studying with the answer key.
Every DPOY since 1985
Where the final model ranked each winner before the season. Green: its favourite won (14). Yellow: the winner was 2nd or 3rd (8). The biggest misses were breakouts: Smart in 2022, Jaren Jackson Jr. in 2023, Mobley in 2025.
Did the favourite win?
Preseason odds for this award only go back to 2015-16, so this is 11 seasons. The model's favourite won 5, the Vegas favourite 3.