The Miraculous Stephen Jackson

Today’s post is another joint effort with Josh Weil (@joshweil) from Everything and Nothing. Josh was great about running the data for this piece and if you’re a fan of High School Win Score numbers, you should check out his latest!

Today’s post comes courtesy of Mark Cuban, who made a very interesting point on a recent post:

I do love the “Miracle of Wages of Wins” that infected Stephen Jackson when he went from a loss leader to all star caliber when he switched teams last season. Im shocked it happened. Not shocked it wasn’t mentioned.

I asked Josh Weil, who has been doing game by game analysis with data from nerdnumbers.com, for a favor and he provided the following graph for S-Jax!

Now, it’s true the Stephen Jackson was absolutely terrible in Milwaukee. It’s also worth noting that Stephen Jackson was 33 last year and players over 30 tend to decline, which is what we’ve been seeing.

In San Antonio, Stephen Jackson did have a resurgence and played the best he had in years. But let’s note a few things:

  • Stephen Jackson was still below average in San Antonio and had more below average games than “All-Star” level ones.
  • A point brought up in Stumbling on Wins is very few coaches seem to have a positive impact on wins. One that does? Greg Popovich, who happens to coach San Antonio!
  • Finally, Stephen Jackson only played 500 regular season minutes with the Spurs and and only 300 in the playoffs. This isn’t the largest period of time.

Here’s something fun to note about Stephen Jackson’s improvement. Out of 20 games, he had four All-Star level games. If we take these games out, guess what? Jackson’s WP48 for his time on the Spurs drops to 0.000 (which is terrible)

Of course, it takes skill to play an All-Star level game right? All of Jackson’s  top games were over 20 minutes. I looked for players that managed to play over 10 games at 20+ minutes and hadn’t managed to put up an All-Star level game. The list only had two names;

Nets fans can certainly feel the pain. The other thing is that even terrible players can have great games. Here’s a fun breakdown:

2011-2012 Players with < 0.000 Wp48 and > 5+ 0.200 WP48 games at 20+ MP

Player Team MP WP48 Great Games
Jordan Crawford Washington Wizards 1,753 -0.029 13
Antawn Jamison Cleveland Cavaliers 2,151 -0.041 11
DeMar DeRozan Toronto Raptors 2,206 -0.007 11
C.J. Miles Utah Jazz 1,145 -0.021 10
Markieff Morris Phoenix Suns 1,227 -0.031 10
Glen Davis Orlando Magic 1,427 -0.023 9
Leandro Barbosa Toronto Raptors 946 -0.021 9
Luis Scola Houston Rockets 2,067 -0.036 9
Patrick Patterson Houston Rockets 1,483 -0.006 9
Boris Diaw Charlotte Bobcats 1,018 -0.011 8
Lamar Odom Dallas Mavericks 1,027 -0.038 7
Michael Beasley Michael Beasley 1,087 -0.038 7
Nick Young Washington Wizards 1,211 -0.049 7
Al Harrington Denver Nuggets 1,761 -0.034 6
Byron Mullens Charlotte Bobcats 1,465 -0.084 6
Corey Maggette Charlotte Bobcats 881 -0.054 6
Josh Howard Utah Jazz 991 -0.011 6
Chris Kaman New Orleans Hornets 1,372 -0.049 5

It turns out there can be downright terrible players that can still put up more than a few great games.

Summing up

Stephen Jackson has been a sub-par player his entire career. He went to a team with one of the few good coaches/systems that improves players. In a short span he was able to put up a few great games that upped his general level. Of course, any player can play great in the NBA. In fact, almost all players that get decent playing time do.

One last point is this. Mark Cuban, in a slightly aggressive tone, impled that Wins Produced is not perfect. Guess what? He’s right! The box score has a ton of data. It can explain ~95% of wins and much of that can be assigned to individual players. What’s more, players are fairly consistent year to year. This doesn’t mean basketball is solved. It just means we’re darn close.

When making any questions about advanced stats, you have to systematically approach the problem and test it. Henry Abbott and Dave Berri showed a great example of this recently:

  • A theory was posited (offensive rebounds impact defense)
  • A criteria was set up (Is there a relationship between offensive rebounds and defensive efficiency?)
  • And an answer was found (no, not really)
In a league with hundreds of players, the simple fact is not all of them will adhere to a model. That’s fine. The question is: how good is the model you’re using? If we focus on things like the eye test and outliers, we’ll never adequately answer that question. I hope Mark Cuban, who is a supporter of the Sloan Analytics Conference, agrees with this take. After all, all of us are looking for the answers in the stats. We just have to ask the right questions.
-Dre

Comments are closed.