Nba nerd

The Good, The Bad, The Lucky, and The Wolves

Today, I fixed something that's been nagging me (and many of our readers) for a bit. If you take a look at any team's stats tables (for example, Minnesota here), you'll notice that there is an "Expected Wins" area in the bottom right. Up until today, this was zero for every team, thanks to a bug in our software that I've only just corrected (yay, day jobs!).

Since the above paragraph would make for a very boring article, I took the liberty of compiling the differences between the total wins produced by each team's players (the "expected wins" value) and each team's actual wins, and then sorting them by "luck". Which gives us a handy table:

Team Wins Produced Actual Wins Lucky Wins
10.4 15.0 4.6
28.0 31.0 3.0
16.9 19.0 2.1
25.1 27.0 1.9
17.3 19.0 1.7
20.3 22.0 1.7
36.6 38.0 1.4
23.6 25.0 1.4
27.7 29.0 1.3
38.8 40.0 1.2
42.0 43.0 1.0
41.3 42.0 0.7
39.5 40.0 0.5
33.8 34.0 0.2
38.0 38.0 --
26.2 26.0 -0.2
10.7 10.0 -0.7
23.9 23.0 -0.9
20.1 19.0 -1.1
27.7 26.0 -1.7
30.1 28.0 -2.1
25.4 23.0 -2.4
35.9 33.0 -2.9
19.9 17.0 -2.9
41.1 38.0 -3.1
35.3 31.0 -4.3
39.0 34.0 -5.0
25.5 20.0 -5.5
27.2 21.0 -6.2
35.0 27.0 -8.0

Of course, what's being captured here isn't really luck at all; it's just the delta between what the wins produced model says about the strength of a team and the actual outcomes. For instance, the Sixers look very lucky because the algortihm looks at MCW's wins produced and averages it over all of his minutes. But what's far more likely is that certain players played very well early on (like MCW) and had a decent/high WP48 at the beginning of the year, and have been awful recently. But unlike wins produced, actual wins can't be negative. In other words, if a player plays like LeBron James for 5 games, and his team wins 5 games, then plays like Bargnani for 35 games, his total wins produced aren't going to stay positive, and the team will look luckier than its WP48, when what really happened wasn't that the team was lucky to win 5 games despite his poor play, but instead that the team was just lucky to get 5 games of solid production out of that player. The difference is subtle but the distinction is important.

Or in other other words, looking at season averages is not going to do a great job of explaining why a team lost any particular game, or even a stretch of a few games. Which, of course, is why we watch the actual games instead of just consulting spreadsheets.

Still, it's nice to think the Timberwolves are "really" a 35-win team and "should be" in 6th place in the West. Or...wait, that's not ****ing nice to think about at all. I'd rather go with my current theory: that playing JJ Barea more than Ricky Rubio in the fourth quarters of close games explains a hell of a lot about why you are 1-15 in close games, and that, Mr. Adelman, if you lament the lack of ball movement, yet you play JJ Barea lots of minutes and hope that the ball moves, I have a definition of insanity that you might want to consider.