Franalysis: A Plausible Way For Iowa To Get Where It Needs To Be On Defense

By houksyndrome on February 20, 2019 at 6:06 pm
go d go
© Jeffrey Becker-USA TODAY Sports
21 Comments

There are four things that determine how good a basketball team is at playing defense:  the rate at which they force turnovers, the rate at which they are called for shooting fouls (free throw rate), the rate at which they concede offensive rebounds and the effective field goal percentage of their opponents.  Effective field goal percentage is equal to 0.5 multiplied by the number of points they have allowed on field goals divided by the number of field goal attempts.  So it factors in three point shooting and weights that appropriately.  The turnover rate, free throw rate, offensive rebounding rate and the effective field goal percentage are collectively known as the four factors.

As of February 17th, Iowa rated very poorly in three of the four factors, defensively.  Iowa is 39th nationally in FT rate but 169th, 167th and 208th in eFG%, TO% and OR%.  The poor eFG% is even more remarkable given that Iowa’s opponents have not shot three pointers very successfully against us (32% accuracy, 62nd nationally).  Iowa’s opponents shoot about 53% (273rd nationally) on two point shots against us and this is entirely the reason for our poor eFG defense.

I decided to play the “what if” game in order to see which factor -- eFG, OR%, or TO% -- was hurting Iowa the most.  I calculated the number of points that would be saved if Iowa was 50th, nationally, in two point defense, OR%, or TO%.  I am not including the details on how I made those calculations in the interest of brevity but I’m happy to discuss this in the comments.  I found that two-point defense would make the biggest difference, saving Iowa about 88 points.  By my calculations, we have given up 1831 points over 1777 possessions.  So 88 fewer points would result in a defense that allows 0.98 points per possession (our current one allows 1.03 PPP).  That would put our defense at about 40th, nationally.  Being 50th in defensive rebounding would result in 49 fewer possessions for our opponents and about 50 fewer points allowed (our opponents score about 1.03 points per possession against us).  Being 50th in turnover generation would result in 55 fewer possessions for our opponents and this would save us 69 points (our opponents score 1.26 points per possession on which they don’t turn the ball over).  So, in terms of what would help us the most, improved FG defense > improved turnover generation > improved defensive rebounding.

This got me wondering if all of the four factors were equally important.  Obviously, at the extremes, they are equally important - if you are called for a foul on 100% of your opponent’s possessions, you are screwed.  If you allow your opponent to rebound 100% of their misses, you are screwed.  If your opponent turns the ball over on 100% of their possessions, they will never score.  If your opponent shoots 0% from the field, they will score very few points.  However, real teams don’t have extreme rates like that.  So, for real teams, which of the four factors have the strongest effect on overall defensive effectiveness?

I analyzed this by examining the correlation between each of the four factors and the number of points allowed per possession.  If you do this kind of correlation analysis over all 350+ D1 teams, you will probably find that everything correlates with everything.  The reason for this is that teams like Duke are better at everything than teams like Savannah State.  Thus, for my analysis, I controlled for this overall quality effect by only looking at teams which were approximately equivalent with each other in overall quality.  Specifically, I analyzed teams that ranked between 50th and 60th in Bart Torvik’s overall efficiency metric for every season from 2014-2019.  So I have 66 data points, each of which represents the performance of a team during a single season (“team seasons”).  The decision to use that specific ranking window was arbitrary.  Full disclosure here: I am not a trained statistician or data scientist.  My methodology might be flawed (and my data set too small).  Still, I noticed some trends here which really caught my eye.  The first is shown in Figure 1.

Figure 1:  Each of the four factors plotted against overall defensive efficiency

Correlation between each of the four factors and overall defense

Field goal defense appears to be a significantly better predictor of overall defensive quality than the other three factors.  It has the highest correlation coefficient with DPPP and has the sharpest slope.  The slope is positive, as it should be; the higher your opponent’s shooting percentage, the more points you will allow per possession.  Turnovers were the next best predictor of overall defensive quality, with a clear negative trend to the data, which makes sense (more turnovers -> fewer points allowed).  In contrast, defensive rebounding didn’t seem to matter at all in the sense that the best defensive rebounding teams in my data set (with opponents collecting 20-25% of their misses) had about the same overall defensive efficiency as the worst defensive rebounding teams in my data set (with opponents collecting 30-35% of their misses).  The best fit line had a very shallow slope with a very weak correlation co-efficient.

Most peculiar of all was the relationship between free throw rate allowed and overall defensive efficiency.  Like defensive rebounding, there was very little correlation (R-squared of 0.01).  Even more curious is that the trend line had a negative slope.  In other words, teams that fouled more had a slightly better overall defensive efficiency.  Why on earth could that be?  Well, it turns out that free throw rate and two-point defense correlate with one another pretty well.  In fact, their correlation coefficient is slightly higher than the correlation coefficient between turnover rate and overall defensive efficiently (R-squared of 0.13 vs. 0.096) for my data set.  In other words, teams that foul more tend to force their opponents into worse shooting performances.  This isn’t terribly surprising:  more aggressive defense leads to more fouls and fewer made baskets.  These data could also suggest that physically aggressive teams get away with a lot of fouls.  Are you thinking who I’m thinking?

So it seems as though improving our FG% defense, and specifically our two-point FG% defense, would generate the most improvement in our defense overall.  How can we improve our two-point defense?  One possible (and highly intuitive) determinant of two-point defense is shot blocking so I examined the relationship between shot blocking and two-point defense for the same 66 team seasons that I used above.

Figure 2:  Shot blocking vs. Two-Point Defense

Figure 2:  Shot blocking vs. 2 Point Defense

The correlation between shot blocking and two point defense is very strong (R-squared=0.44).  Just look at the data in figure 2 for yourself.  As teams block a larger percentage of their opponents’ shots, their opponents’ two-point field goal percentage drops significantly.  In my data set, every single one of the teams that allowed their opponents to make above 50% of their two-pointers blocked less than 10% of their opponents’ two point shot attempts.  Similarly, all of the teams who held their opponents to 45% or less on two pointers blocked at least 11.4% of their opponents’ twos.  I have played (and watched) enough basketball to know that good shot blockers are very difficult to score on.  Not only do they block shots but you have to alter all your shots in order to avoid blocks which also hurts your accuracy.  Still, I was stunned by the clarity of the relationship between the two stats.  I expected to have to analyze way more than 66 team seasons for this relationship to show up.

How does this relate to Iowa?  Over the course of the year, Iowa has blocked about 9.6% of their opponents’ two point attempts (148th nationally).  In Big Ten games, that number falls to 8.7% which is 12th in the conference.  Nicholas Baer is the primary reason that those numbers aren’t even worse.  Bart Torvik estimates that Baer blocks 6.2% of his opponents’ two point attempts when he is on the court, which puts Baer in the 98th percentile of shot blockers, nationally.  Kriener is decent at blocking shots too (3.5% / 78th percentile).  Cook and Garza, though, not so much.

Figure 3:  The shot blocking rate and fouling rate for the major front court players in the Big Ten

Figure 3:  Shot blocking rates and fouling rates for major Big Ten front court players

 

 

 

 

 

 

 

 

 

 

 

I generated a list of all the major front court players (4s and 5s) for every Big Ten team (37 players) and ranked them according to their shot blocking rates (Figure 3, above).  Tyler Cook ranked 34th on that list (again, 37 players) with a 2% shot blocking rate.  Garza’s 2.9% shot blocking rate puts him 27th on that list.  So our starting PF and C, who combine for about 50 minutes a game, don’t block very many shots relative to their peers.  Garza is not the springiest dude in the world, but he blocked 4.6% of his opponents’ shots last season, so I think there is room for improvement with him.  4.6% would help us out a lot.  Why doesn’t Cook block more shots, though?  He obviously has the athleticism to be a great rim protector.  I have a theory on that.  Figure 3 (right) shows Big Ten front court players ranked according to how many fouls they commit.  As you can see, Cook commits the fewest fouls per forty minutes of any big man in the conference.  That right there is my explanation for why Cook blocks so few shots - concern over fouls.

Figure 4:  Shot blocking rates vs. fouling rates for major Big Ten front court players

Figure 4:  Shot blocking rate vs. fouling rate for major Big Ten front court players

It makes logical sense for us to be concerned about Cook picking up fouls.  Still, I think Cook could be a bit more aggressive without getting in major foul trouble (not to mention that a good deal of his fouls seem to be charges).  The correlation between blocked shot rate and fouling rate is actually pretty weak (R-squared=0.02, Figure 4) and there are a number of good shot blockers who don’t commit a ton of fouls.  One good example is Nicholas Baer who combines a 6.2% block rate with a low fouling rate (3 per 40 minutes).  Ethan Happ blocks 4.2% of his opponents’ two pointers and only commits 3.3 fouls per forty minutes.  There are other examples, too.

If Cook blocked 4% of his opponents’ two pointers and Garza got back to 4.6%, all other things being equal, our shot blocking rate as a team would improve to 12%.  In my 66 team season cohort, no team blocked 12+% of their opponents’ shots and allowed their opponent to shoot over 47.1% on two pointers.  So, if Cook and Garza were to block shots at the above rates, my calculations predict that Iowa’s two point defense would improve to 47% (or even better).  This would get Iowa’s two point defense into the top fifty, nationally, and would get Iowa’s overall defense into the top 40-50.  That top 40-50 overall projection makes some assumptions that might not hold perfectly (especially our team’s good defensive FT rate holding up) but I think it’s pretty clear that this defense would get dramatically better if our big men blocked/contested more shots.  I believe they could accomplish this without their fouling rates spiking to the point that they wouldn’t be on the court enough to influence games on the offensive end.

Congratulations.  You made it to the end of my article.  You now hold a Master’s Degree in Basketball Analysis from GIA University*.

*GIA University is not a real thing.

**Also, most of these data came from Bart Torvik’s awesome (and free) website.  The rest came from ESPN.  So major hat tip to Bart!

21 Comments
View 21 Comments