courtesy Five Thirty Eight in case you like numbers;
[url=
https://fivethirtyeight.com/features/weve-made-a-slight-correction-to-our-nba-model-sorry-luka/?addata=espnallas:index]
We've Made A Slight Correction To Our NBA Model. (Sorry, Luka!)[/url]
By
FiveThirtyEightMAY 17, 2022, AT 1:32 PM
Going into the playoffs,
our NBA model wasn't particularly high on the Dallas Mavericks' chances of making a deep run. We
gave Dallas only a 37 percent chance of beating the Utah Jazz in the first round and a 12 percent chance of making the Western Conference finals. But that was before Luka Doni returned from injury and then
went nuclear in the playoffs, averaging 31.5 points, 10.1 rebounds, 6.6 assists and 1.9 steals per game with a 58.7 true shooting percentage and an absurd +14.9
RAPTOR plus/minus rating. Now the Mavs are on the cusp of an NBA Finals appearance, and as of this morning our model was very high on Dallas's chances.
A little bit
too high, as it turns out.
As a quick refresher,
our RAPTOR projections incorporate an adjustment for playoff performance based on a player's career RAPTOR ratings in the playoffs versus the regular season, with an emphasis on more recent seasons. But after digging around earlier today, we found that we had been implementing this in a clumsy way for our real-time projections, such that they were adjusting more than we intended for 2022 playoff performance.
This was most noticeable in the case of Doni, who is having a historically great postseason. Before fixing the bug in our playoff adjustment, Doni was appearing as nearly a +13 points-per-100-possessions player, relative to league average. If that was indeed his true talent, he would be on par with
Michael Jordan or LeBron James at their peaks and he has played even better than that so far in the playoffs. But a player's RAPTOR playoff adjustment should evolve very slowly over time, since we operate under the strong belief that most players play roughly the same in the playoffs as they do in the regular season, with a few exceptions. With only
857 career playoff minutes under his belt, Doni hasn't quite played at this level in the postseason long enough to justify a LeBron-like adjustment, much less a playoff adjustment several times larger than any player in NBA history.
Doni's playoff adjustment is now a much more reasonable +0.6 on top of his already great regular-season rating. That means the model is still comparatively bullish on Luka and the Mavs (we give them a 54 percent chance to make the Finals), just not as much as it was before we discovered this bug. Though if Doni keeps playing as well as he did against the Jazz and Suns, no playoff adjustment will really matter in the end.