Franalysis: Player Efficiency Data from First Five Games

By houksyndrome on November 28, 2019 at 12:00 pm
go joe go
© Jeffrey Becker-USA TODAY Sports

Hi folks!  We’re back for another season of Franalysis.  Happy Thanksgiving!  I’m thankful for Megan Gustafson, AJ Epenesa, Spencer Lee, and pivot tables.  The men’s basketball team is facing adversity right now but, I assure you, no matter what happens this season, I’ll be watching all of the games with an open Excel spreadsheet. 

Just to quickly remind you exactly what I keep track of, I record the result of every possession for each team during every Hawkeye game, as well as the player combination that we had on the floor and the type of defense that we played on every possession.  Over time, this will allow us to examine systematic differences in the effectiveness of particular defensive schemes and lineup combinations. It will also give us another metric to evaluate individual players (in addition to the conventional box score-based metrics).

In this article, I will be focusing on how the team performs when a particular player is on the floor.   Does the offense get better or worse?  Does the defense get better or worse?  In basketball analytics, what I am measuring is something akin to “real plus minus.”  The idea behind this statistic is to capture things that don’t show up in the box score.  It is not a perfect metric.  Our starters are usually playing against the other team’s starters, whereas our bench players are going against the other team’s bench players (or the other team’s tired starters).  This can distort the plus minus in favor of bench players.  Also, if one player tends to play with a specific set of teammates then that can also distort his plus minus (bench players play with other bench players etc.).  As the sample size gets larger, I can tease apart the effect of teammates.  I can’t do much about opponent quality, though, without doing possession tracking for all of our opponents' games (i.e. hundreds of games).  Currently, I have a small sample size (five games, four of which are against weak competition).  The good (and bad) news is that from here on out, we will be playing top-notch competition.  So I will be updating my data set and writing more articles like this one as the season goes along and we will soon find out if these early trends continue to hold up.

Fran’s teams were pretty solid, defensively, when we had Adam Woodbury, Mike Gesell, and Anthony Clemmons.  Since then, we haven’t been any good defensively.  There has been much discussion over why that is and many hypotheses have been put forth; however, a consistent explanation has been a lack of perimeter quickness.  I have wanted to look into this for a long time; however, I couldn’t because we didn’t have any particularly quick players.  Now we have Joe Toussaint, though, and we can compare how the team does when Toussaint is on the floor versus when he is off the floor.  Let’s start with some tables of how the team does when each of our players on the floor (one for each game and a grand total at the end).

C. McCaffery 1.2 0.87
Fredrick 1.33 0.84
Wieskamp 1.16 0.85
Kriener 1.31 0.71
Garza 1.2 0.8
Nunge 1.25 0.82
Evelyn 1.26 0.78
P. McCaffery 1.23 0.83
Bohannon 1.3 0.83
Toussaint 1.18 0.85
Pemsl --- ---


C. McCaffery 0.86 1.16
Fredrick 0.92 1.21
Wieskamp 0.91 1.13
Kriener 1.06 1.18
Garza 1 1.14
Nunge 0.79 1.47
Evelyn 1.07 1.21
P. McCaffery 1 1.26
Bohannon 1.03 1.31
Toussaint 0.93 0.93
Pemsl 0.93 0.89
Oral Roberts
C. McCaffery 1.33 1.04
Fredrick 1.38 0.95
Wieskamp 1.28 0.91
Kriener 1.08 1.21
Garza 1.31 0.96
Nunge 1.27 0.94
Evelyn 0.95 1.21
P. McCaffery --- ---
Bohannon 1 1.1
Toussaint 0.92 1.25
Pemsl 0.95 1.21
North Florida
C. McCaffery 1.19 1.17
Fredrick 1.36 1.06
Wieskamp 1.29 0.93
Kriener 1.17 0.92
Garza 1.27 1.12
Nunge 1.28 1.13
Evelyn 1.25 1.18
P. McCaffery --- ---
Bohannon 1.3 0.84
Toussaint 1.23 1.03
Pemsl 1.3 0.92
Cal Poly
C. McCaffery 1.26 0.81
Fredrick 1.29 0.88
Wieskamp 1.39 0.91
Kriener 1.35 1.04
Garza 1.4 0.89
Nunge 1.14 0.75
Evelyn 1.04 0.64
P. McCaffery --- ---
Bohannon --- ---
Toussaint 1.22 0.67
Pemsl 1.21 0.53
Total through five games
C. McCaffery 1.16 1.0
Fredrick 1.24 1.0
Wieskamp 1.2 0.95
Kriener 1.19 1.01
Garza 1.23 0.98
Nunge 1.19 1.0
Evelyn 1.12 0.98
P. McCaffery 1.14 1.0
Bohannon 1.15 1.03
Toussaint 1.12 0.90
Pemsl 1.1 0.85

I want to reiterate that these data have caveats, as discussed above, and this is a small sample size but, wow, the team’s defensive efficiency when Toussaint is on the floor really stands out doesn’t it?  Overall, we have scored 1.18 points per possession and allowed 0.97 points per possession.  When Toussaint is on the floor, we score 1.12 and allow 0.9 points per possession.  When Toussaint is not playing, we score 1.21 and allow 1.02 points per possession.  We have been allowing 0.12 fewer points per possession when Toussaint is on the floor.  That is a very big difference.  If you could magically deduct 0.12 PPP from our defense last year, we would have been 5th, nationally, in defense and one of the very best teams in the country overall.  Again, small sample size, weak competition, playing against the other team’s bench, these caveats all apply.  But still.  There seems to be something here, defensively, with Toussaint.

Another thing that stands out from those data points, at least regarding Toussaint, is that the offense is less effective when he is on the floor (1.12 PPP with him versus 1.21 PPP without).  I think this is due to him rarely being on the floor with Garza.  Garza has been on the floor for about 2/3 of our possessions overall.  We score 1.23 PPP with Garza and 1.06 PPP without him.  Only about 25% of Toussaint’s possessions also occurred with Garza on the floor and we scored 1.23 PPP on those possessions, equivalent to Garza’s overall rate.  When Toussaint is on the floor without Garza, we score 1.09 PPP.  When neither Toussaint nor Garza are out there, we score a paltry 0.85 PPP.  To me, these data collectively suggest that Toussaint’s depressed offensive plus minus is an artifact of him not playing with Garza very much.

One last thing that jumped out was how effective we’ve been defensively when Pemsl is on the floor.  We have allowed 0.85 PPP with him and 1.02 without.  The first thing that jumped into my head when I saw that statistic was “small sample size fluke / artifact”; however, the more I think about it, the more I think it might have some validity.  I doubt it holds up at 0.17 PPP difference over the course of the year (0.17 is enormous), but I do think that Pemsl adds something defensively that the rest of our bigs lack.  He has much more lateral agility than Garza, Kriener, or Nunge and, since we often switch on perimeter screens, that ends up being critically important.  Cook, despite being a non-factor as a shot blocker, still had a positive defensive rating according to my metric last season.  We were better defensively when he was on the floor.  I think his ability to switch on screens was a reason why.  Baer could also do that.  This is my working theory for why my defensive metric loves Pemsl so much.  

We will find out whether these trends hold up over the course of the season. 

View 4 Comments