Rocketscience: Analytics Teams vs. Non-Analytics Teams

The Rockets have fared very well this season against below .500 teams.  They have sported a record of 14-6, to be exact, which is right up there with the top-tier teams in the league.   Instead, our problem this season has been against the good teams, with an unimpressive 10-22 mark.   I also noticed that Portland, a team with a similar bias towards advanced basketball analytics, has a similar problem (7-17 vs. above .500, 20-7 vs. below .500), which got me thinking – maybe that whole “too good to be true” feeling has something to do with how the most common metrics normalize data.   Most of the metrics are averages, perhaps correcting for playing time, teammates on the court, and even time left in the game or shot clock, but how much attention is being paid to whether or not a player elevates his game versus better opponents?

The answer to that question is beyond this post, but we can at least take a look at the team-level for wins and losses.  According to nbastuffer.com, there are 18 teams in the league with an analytics department, or at least affiliation with analytics consultants.  Without delving too deeply into the depth of each team’s analytical reliance, I compared the records of “analytics” teams vs. “non-analytics” teams with regard to above .500 and below .500 opponents.  Here is the full breakdown:

And here are the winning percentages:

I was hoping for a defining relationship that would be similar to the Houston/Portland problem, but found a somewhat different trend.*  It looks like teams that incorporate some form of advanced analytics have better records – both against good teams and bad teams.  On top of that, I think an argument could be made that the top teams are winning despite that handicap.  The Bulls are loaded with talent and have excellent coaching in Thibodeau.  New Orleans would be nowhere without Chris Paul.   Finally, the Jazz have one of the best coaches in the league and the hardest offense to guard in the league.

Of course, you could make the coaching/talent argument with some of the top analytical teams too, but of San Antonio, Boston, LA, and Dallas, how much of their talent was a result of their draft position and how much was a result of a keener eye?  Besides Duncan, the Spurs’ best players are Ginobili (57th overall), Parker (28th overall), and Hill (26th overall).   The Celtics found Rondo at 21st overall, LA got Kobe at 13th, and Dallas traded a Tractor Trailer for Dirk Nowitzki.

I’m not sure what else to conclude from this, other than 1) the Rockets’ woes against above .500 teams should not be attributed to the team’s penchant for advanced analytics, and 2) getting poor draft position is no excuse for losing.

*One way to take a better look at these records would be to expand the data back to include a couple more seasons, and to make some adjustment for the team’s overall record (because, for example, the Spurs are going to have a good record against both good and bad teams).

Written by Ben Heller, ‘Rocketscience’ is a column devoted to basketball analytics.  Ben Heller can be contacted at heller.benjaminj@gmail.com.






in essays
  • Dean Oliver left the Nuggets recently.

  • DT

    Sometimes when you get buried in the numbers, you miss the simplest ones….24th in the league in Defense. We have 1.5 defenders and 12.5 offensive players playing o’le defense so they can get the ball back. We lost our defensive coach (thanks Chicago/Boston) and our current mentality is to go 1:1 with everyone. The Texans did the same thing this past season…

  • Easy

    Shouldn’t you look at the differential between a team’s record against below .500 opponents and it’s overall record, and look at the average of these differetials among analytic teams and non-analytic teams?

    Non-analytic: 15.16%
    Analytic: 14.31%

    These numbers seems to show that the trend is the opposite of the Rockets and the Blazers.

  • Easy

    Shouldn’t you look at the differential between a team’s record against below .500 opponents and it’s overall record, and look at the average of these differetials among analytic teams and non-analytic teams?

    Non-analytic: 15.16%
    Analytic: 14.31%

    These numbers seems to show that the trend is the opposite of the Rockets and the Blazers.

  • Easy

    BTW, I don’t know how significant the notion of “stepping up against strong opponents” is. To me, that is the same thing as “playing down to the weak opponents.”

    Perhaps the common denominator of the Rockets and the Blazers is that they both are talent-depleted by injuries to their stars. Their records shows that they beat the teams they should beat and cannot beat the good teams because of lack of talent. It seems that it indicate discipline. You play hard every game and you win the games you should win. But talent is still needed for winning against good opponents.

  • ben heller

    I thought about different ways of accounting for a team’s total record when looking at splits, but any time you do that it takes away from the big picture, which is that teams with analytics departments had better records either way. Interesting, though, that the spread is more narrow for the analytics teams.

  • ben heller

    Thanks. Yeah it mentions that on nbastuffer.com too, but I figured his impression is still on the team until they blow it up.

  • Anonymous

    I don’t think you can draw any conclusion that supports your argument with less than a 10% difference on stats..

    This tells me that analytics may be overrated, that all they do is quantify that which is just as easily analyzed with a good trained eye.

  • I would be interested in seeing numbers that are adjusted based on the record of the opposition. I guess the average team in the NBA would be slightly below .500. So weigh the output/efficiency/whatever numbers based on the record.

    Should be pretty easy to do. Does something like this exist already?

  • Anon

    2010-11 Analytics teams vs. Non-Analytics teams (221-149 for .597 record)
    – Only Chicago and Atlanta had better than .500 record against Analytics teams

    2009-10 Analytics teams vs. Non-Analytics teams (429-199 for .683 record)
    – Only Charlotte, Atlanta, and Utah had better than .500 record against Analytics teams

    Note: New Orleans was added to list of Analytics teams for 2010-11. NJ was taken out of 2009-10 Analytics teams (Lee started in 2010-11).

  • ASK
  • Pingback: Teams With Advanced Stat Analysts Winning More NBA Games in 2010-11 | Load O' Bull()

Follow Red94 for occasional rants, musings, and all new post updates
Read previous post:
Post-game interviews: 2.5.11
Close