Assessing Team Chemistry

I found an interesting blog entry as I was doing some reading tonight.  Jamie Merchant has an entry at his blog titled Numeranda where he takes a look at a team’s adjusted plus/minus stats and compares it with the sum of the individual players on each unit’s adjusted plus/minus stats.  Here is his entry.  The hypothesis here is that 5 man units that outperformed collectively the sum of their individual APMs have good chemistry, whereas 5 man units that underperformed collectively the sum of their individual APMs may not have great chemistry.  Note that this is a study on team chemistry, using APM as a means of evaluating this and putting a numerical value on it, rather than the quality of the team.  This can be seen in the fact that some of the higher rated teams relative to chemistry, using this model, are actually terrible teams that didn’t win a lot of games. 

In looking at the numbers to relate this to the Hawks, the three units the Hawks used the most this past season have negative chemistry ratings.  Note that the chemistry rating is the difference between the team’s APM vs the sum of the individual players’ APM.   The Hawks’ three most used 5 man units are:

  1. Mike Bibby/Joe Johnson/Marvin Williams/Josh Smith/Al Horford
  2. Mike Bibby/Jamal Crawford/Joe Johnson/Josh Smith/Al Horford
  3. Jamal Crawford/Joe Johnson/Marvin Williams/Josh Smith/Al Horford

This is how these units rated relative to the chemistry rating, with team APM and sum APM included:

  1. Bibby/Crawford/Johnson/Smith/Horford – “chemistry” -2.51 (APM of 7.22 – sum APM of 9.73)
  2. Bibby/Johnson/Williams/Smith/Horford – “chemistry” -3.05 (APM of 3.76 – sum APM of 6.81)
  3. Crawford/Johnson/Williams/Smith/Horford – “chemistry” -4.25 (APM of 7.85 – sum APM of 12.1)

It is difficult to make any conclusions on this date at this point because the sample sizes are too small.  For discussion purposes though, where does the source of the relative lack of chemistry lie? 

Six players on the Hawks were used the most this past season.  Here are their individual APMs:

  • Mike Bibby -2.84
  • Joe Johnson -5.15
  • Marvin Williams -0.47
  • Josh Smith 4.88
  • Al Horford 10.39
  • Jamal Crawford 2.45

I will admit that I don’t quite understand how the APM is derived.  I do know that it takes into account the quality of the teammates the players are on the court with as well as the quality of the player’s that play the same position on this team.  It also takes into account the quality of the opponent, or player matched up with.  That could explain why Joe Johnson’s APM is the most negative of the bunch even though he is generally considered to be the best player on the team.  Joe is typically matched up with the oppositions best offensive player when he is on defense, and he typically draws the oppositions best defensive player if not a double team when Joe is on offense.  Conversely, Al Horford is typically matched up with the oppositions center, and for the most part, there are very few centers in the league that can give him trouble either offensively or defensively.  Hence a likely reason why Al Horford has the most positive APM on the team.

So, back to the question at hand.  Why does it seem like the Hawks top three units lack chemistry relative to this model?  How much of this is a situation where the team was simply put together poorly without taking into account how each player would fit with each other, and how much of this comes down to coaching?

My initial thoughts on this is that the lack of chemistry has more to do with coaching than anything.  The Hawks under Mike Woodson were maddeningly inconsistent defensively.  A lot of that had to do with a lack of effort from certain players on the defensive end, but I think a lot of it had to do with a complicated switching style of defense that Mike Woodson used.  As a result of the switching, teams were able to create mismatches rather easily against the Hawks.  Offensively, there was so much isolation with Joe Johnson and Jamal Crawford when they weren’t on the floor with each other that I feel it diminished the overall effectiveness of everyone on the team.

Since the Hawks are keeping the same basic line up with the addition of utilizing Jeff Teague more and changing their style, it will be interesting to compare the numbers after this coming season.  How much of an effect will Larry Drew’s more up tempo, motion offense and man to man defense have on the individual players and the chemistry of the team?  Larry Drew is of the opinion that his new system will make things easier on Joe Johnson, thus possibly making him a more efficient and productive player overall. 

A couple of other things I want to point out from this article:

 The Cavs, although a good team (they won 66 61 games after all), used line-ups like a bad team.  They never found that one line-up that was their absolute best, or if they did, they didn’t use it very frequently.  Scanning through the line-ups they did use, you see some promising candidates: Williams, Parker, James, Varejao and Ilgauskas rated +23.87 in APM (but appeared together for only 96 regular season minutes); Williams, Parker, James, Jamison and Varejao rated +25.27 (67 minutes); Gibson, West, James, Varejao and Ilgauskas rated 49.36 (!) but played just 42 minutes together.  Standard-errors wreak havoc with these numbers at such low minute totals, but it would have been interesting to see the Cavs devote more minutes to these line-ups.

The Cavs may actually be a great example of bad coaching not really having an effect in the regular season but having a profound effect in the post season.

Their line-up problems continued in the playoffs, where their most-used line-up (Williams, Parker, James, Jamison, O’Neal) actually had a negative APM (-.14 in 166 minutes), and their next most-used line-up (Williams, Parker, James, Jamison, Varejao) was outstanding (+24.73) but was only used about one third as much (55 minutes).

The head coach is responsible for knowing how effective his line ups are and putting the more effective ones on the floor.  When looking at the Hawks, their two most effective line ups during the regular season were:

  1. Jamal Crawford/Maurice Evans/Marvin Williams/Josh Smith/Al Horford – 55.98 APM
  2. Jamal Crawford/Joe Johnson/Maurice Evans/Al Horford/Zaza Pachulia – 41.03 APM

These two units only played 45.1 and 35.38 minutes during the regular season.  In the playoffs, the top four most effective units for the Hawks:

  1. Jamal Crawford/Joe Johnson/Marvin Williams/Al Horford/Zaza Pachulia – 31.54 APM
  2. Mike Bibby/Joe Johnson/Marvin Williams/Josh Smith/Zaza Pachulia – 22.69 APM
  3. Jamal Crawford/Joe Johnson/Maurice Evans/Al Horford/Zaza Pachulia – 18.40 APM
  4. Jamal Crawford/Maurice Evans/Marvin Williams/Josh Smith/Al Horford – 13.35 APM

Once again it is a small sample size.  However, these four units played respectively 7.48, 4.88, 11.95, and 10.32 minutes in the playoffs out of a total of 528 minutes played.  The most used line up in the post season was Mike Bibby/Joe Johnson/Marvin Williams/Josh Smith/Al Horford with a -4.43 APM.  It is very difficult to say that Atlanta would have had a better chance of winning more in the playoffs had they utilized the first line up more.

The last thing I want to point out is this.

There is a certain 350-pound elephant in the room here and I might as well mention it now: Shaquille O’Neal.  O’Neal caused the Cavs serious problems last year, not so much because of how he played (although his APM was an abysmal -8.44), but because his presence kept another key player off the court: Anderson Varejao.  The APM numbers for the Cavs shoot up whenever Varejao’s on the court (and Varejao’s own APM was a solid 2.61, good for 20th in the league), but unfortunately Varejao’s game did not mesh well with O’Neal’s.  Both were poor shooters, and so, the Cavs coaching staff must have reasoned, it was ill-advised to play them both at the same time.  That’s probably true, but the coaches didn’t go the extra step and reach the obvious conclusion: Varejao is better than Shaquille O’Neal and if we have to choose between them, we should choose Varejao.  Varejao doesn’t quite have the the same rep as Shaq, though, and apparently the coaches could never bring themselves to make the difficult decision to bench a Hall of Fame player in favor of a goofy-looking Brazillian who has never averaged more than nine points a game.

This post has gotten a little off the original topic of chemistry, about which I have more to say (regression analysis ahead!), so I’ll wrap it up.  One last comment on the Cavs though: the Cavs probably tinkered a bit too much with their chemistry by bringing in Shaquille O’Neal.  For a team built for speed, defense and fast-breaking, O’Neal simply did not fit.  Worst of all, he displaced someone who did.  Even though Varejao played more minutes per game (28.5) than O’Neal (23.4), the fact that O’Neal was in the starting line-up meant that Varejao was left out of the game when the other team’s best players were on the court and went in when his own team’s best players were off the court.  The inefficiency of the Cavs’ plan is reflected in their APM numbers, especially the amazing fact that their most-used line-up was significantly negative (-7.13, as mentioned above).

Shaq had a tremendously negative effect on the Cavaliers from relative to his on the court play and relative to chemistry.  To me, this further cements why bringing Shaq to Atlanta would be a bad idea.  Shaq is a shell of his former self, and his slow, out of shape body simply cannot keep up with today’s NBA post players.  The fact that the Hawks play well when Zaza is on the floor further suggests that their resources could be better spent on other positions.

Want more from Soaring Down South?  
Subscribe to FanSided Daily for your morning fix. Enter your email and stay in the know.
  • tim

    I love your ending comment (that Shaq is not the crown jewel we should be seeking) but I do have a problem with APM. I love that Basketball is developing my sophisticated stats other than raw production numbers, but APM seems like an over done version of +/-. Advanced stats work great in baseball because of the player’s lack of reliance on their teammates to perform. But true team sports like basketball have so much going into them that advanced stats, in my opinion, are only valid within the context of a single team. That’s why I think the Shaq/Verijoke argument is valid.

    I also find it interesting that some of the negative Hawks lineups mentioned had some of the best +/- in the league at different points of the season.

    • keithbox

      There are definitely some flaws with the stats, but I think stats like APM that try to take into account the effect of teammates and opponents, though I do not know exactly how they do it, and stats like Win Shares that measure a player’s impact on winning are better measures to use than your standard points scored per game and such.

      • tim

        I totally agree. And I had never heard of APM so thanks for the knowledge drop. It’s a really interesting idea, trying to measure chemistry. I wonder if chemistry ratings are higher for groups that have played together longer. It would also be cool to see how the Hawks chemistry rating has changed over the course of the last five seasons.

TEAMFeed More Hawks news from the Fansided Network

Hot on the Web From golf.com