|
After spending the last two days looking forward by considering the most extreme in-season projections combining performance to date with SCHOENE's preseason benchmarks--most improved and biggest drops--today it's time to look back. I did the same exercise at this point of the 2011-12 season (in January due to the late post-lockout start).
Using Dougstats.com's player stats by day, I was able to reconstruct approximately how each player on both lists performed after the articles were published, giving us an idea how accurate the in-season method proved. First, here are last year's early overachievers. Pre is SCHOENE's preseason projection, "In" is the in-season update and "Rem" is the player's winning percentage the remainder of the season. "Min" is minutes played thereafter; note that one player, Damion James, did not play at all due to injury.
Player Pre In Rem Min
-----------------------------------------------
Daniel Green .470 .585 .558 1218
Vladimir Radmanovic .436 .543 .448 442
Spencer Hawes .462 .566 .507 596
MarShon Brooks .386 .489 .345 1241
Nikola Vucevic .410 .505 .425 637
Paul Millsap .556 .646 .611 1682
Joe Johnson .483 .569 .568 1542
Kawhi Leonard .410 .496 .585 1147
Isaiah Thomas .392 .478 .560 1447
Chandler Parsons .398 .481 .465 1492
-----------------------------------------------
Average .440 .536 .507 1144.4
And the slow starters:
Player Pre In Rem Min
-----------------------------------------------
Shawne Williams .524 .389 .381 319
Jamaal Magloire .355 .254 .285 208
Toney Douglas .577 .476 .350 313
Patrick Patterson .520 .428 .358 1226
Austin Daye .508 .417 .378 463
Damion James .367 .297 - 0
Landry Fields .520 .450 .440 1435
Mehmet Okur .515 .446 .322 110
Tyrus Thomas .555 .496 .343 799
Glen Davis .401 .343 .408 1089
-----------------------------------------------
Average .515 .431 .372 595.9
What stands out right away is the enormous difference between the two lists. The players who started fast faded thereafter, even more so than the in-season projection suggested--though they still outplayed their SCHOENE projection entering the year. By contrast, the poor starters played far worse than their in-season projections and were nowhere close to preseason expectations. Overall, the in-season update removes about a third of the average error in projecting the remainder of the season, justifying the effort.
One clue as to why these groups could fare so differently the rest of the season comes from the minutes column. Naturally, the overachievers got more opportunity to continue to play, even when they declined. Meanwhile, the poor play of the second group sent them to the bench, limiting their opportunity to prove the early play was a fluke.
There does appear to be more to the issue than that, however. Even among the four players who saw at least 750 minutes of action the rest of the way, only Glen Davis matched or beat his updated projection. (In fact, Davis ended up playing better than SCHOENE predicted at the start of the year.) Landry Fields came close to his in-season projection, while Patrick Patterson and Tyrus Thomas maintained their poor level of play all season. In Thomas' case, conditioning was the culprit, while Patterson's regression was largely mysterious, especially considering that he's played well as a starter this season.
Playing well in the early going proved slightly less telling in terms of long-term prospects. While fast starts portended that rookies Kawhi Leonard, Chandler Parsons and Isaiah Thomas would make the All-Rookie Team, fellow first-year players MarShon Brooks and Nikola Vucevic fell off dramatically. Brooks was even worse than his college translation suggested, and is now struggling to get playing time in Brooklyn.
The better non-rookies were projected entering the season, the more they tended to maintain their strong early play. That's true of veterans Joe Johnson and Paul Millsap, as well as emerging Daniel Green. Spencer Hawes and Vladimir Radmanovic largely returned to their lesser early caliber of play.
The one factor I'd like to study more in the future is how the shape of a player's stat line affects the validity of early-season performance. Shooting, especially from beyond the arc, tends to be more volatile than statistics where the denominator is total plays on the floor (assists, steals, blocks, usage) or all missed shots (rebounding). Of course, that also means we should be more skeptical of preseason projections for shooting, so in practice the in-season updates tend to weigh all factors about the same. I'm not sure whether that would hold up to increased scrutiny.
Kevin Pelton is an author of Basketball Prospectus.
You can contact Kevin by clicking here or click here to see Kevin's other articles.
| 
Tough Acts to Follow (11/28)
0 comments have been left for this article. 
