Measuring the Impact of Iowa's Awful Defense

By Matthew Lundeen on February 21, 2018 at 1:00 pm
Iowa's defense has been horrendous this year, but how many wins has it cost them?
© Jeffrey Becker-USA TODAY Sports
9 Comments

Iowa's defense has been horrendous this year, but how many wins has it cost them?

A couple of weeks ago, tucked away at the end of the Franalysis post on Iowa's win against Minnesota, I wrote a handful of paragraphs about the ridiculous disparity between Iowa's offense and defense. At the time -- exactly three weeks ago -- the Hawkeyes were ranked 29th in the country in offense, according to Kenpom's adjusted offensive efficiency, and 239th in adjusted defensive efficiency. At the time, I noted the small company of Power Five teams that Iowa was in over the last 17 seasons, and how those teams generally hadn't been very successful with a good offense and a defense ranked north of 200. Out of that small sample, two teams finished with an overall winning record (something not possible for this Iowa team), and only one team (2015 Indiana) had made the NCAA Tournament. 

The moral of that story was that it's hard to win with a defense this bad. And, really, a Power Five team that isn't breaking in a new coach shouldn't have a defense this bad. Yet here we are three weeks later and (as of me writing this on Tuesday) the defense is now ranked 250th.

But I don't want to rehash the problems with this year's team. We've talked about the defense plenty this year, and with two regular season games left, it is what it is. Instead, when I was doing the research to figure out which teams had been so extremely good on offense and extremely not on defense, I started to get curious about how much even a marginal improvement in defense would have helped this Iowa team. I mean, 250th in the country is really bad, but what if they improved to 225th? What about 200? If the offense was able to maintain the same output (currently 25th in the nation), what would be the value of having a defense that was simply bad instead of burn this season down and salt the earth awful? 

In order to answer that, I looked at the adjusted defensive efficiency for every team at certain rankings. Here is a table:

Rank 250 225 200 175 150 125 100
Adj. DE 109.8 108.3 107.2 105.9 104.9 103.5 102.3
Expected W% 0.668 0.702 0.726 0.753 0.773 0.799 0.820

Note: Much of the math I am about to do can be explained pretty well here.

If you look at the adjusted defensive efficiencies, you can see that for every 25 ranking spots a team improves, their adjusted defensive efficiency gets better by roughly one point per 100 possessions. Using these numbers and assuming that Iowa's offensive output would remain static (an iffy assumption, I'll admit, given that they were ranked in the 60's earlier this season), I went ahead and calculated Iowa's expected winning percentage if their defense were ranked accordingly, and that is what you see in the bottom row of the table.

(Kenpom used to rank teams based on an expected win percentage that he calculated from their adjusted offensive and defensive efficiencies. That changed a couple of years ago, though, and he now uses adjusted efficiency margin. I'm using the former method, however, as it works fine for the kind of estimates I'm trying to make in this post. The win probabilities still come out to about the same. For instance, Kenpom currently gives Iowa a 40% and 57% chance of beating Minnesota and Northwestern. This method gives Iowa a 41% and 54% chance.)

After calculating Iowa's expected win percentage, I calculated the same thing for the rest of the Big Ten. And once that was done, I plugged them in to Iowa's schedule and ran a thousand simulations of how many games Iowa would be expected to win based on how good their defense was. Here are the results in table form:

wins

Here they are in graphical form:

And here are the average conference wins at each defensive ranking:

Def. Rank Average Conference Wins Average Added Win Prob. Per Game (Compared to #250)
250 5.8 N/A
225 6.4 3.2%
200 6.9 5.7%
175 7.5 8.7%
150 7.8 11.0%
125 8.4 14.5%
100 9.1 17.5%

Obviously, we all know that Iowa would have won more games this season had they been better on defense. But, by using this model, we can estimate how many more wins they would have at each ranking marker. We can also estimate how much better Iowa's odds to win a game would be, on average, based on how improved the defense was. 

Before we discuss the results, I want to point out a few things about the average win estimate for this current team with a 250th ranked defense. As you know, Iowa has all of three Big Ten wins right now, and with only two regular season games remaining, it's clear that they are not going to reach that average win mark of approximately six. It is worth noting that Iowa is ranked 340th in the country in Kenpom's luck metric. That could be a sign that Iowa has under performed this year, or it could also mean that his model doesn't capture this team very well. (A discussion for another day, considering that only two of Fran's Iowa teams have ever finished better than 200 in the country in "luck", and four of them have been ranked north of 300.)It's also worth noting that the 5.8 average wins is based on the assumption that Iowa played their entire conference schedule with the 25th ranked offense and the 250th ranked defense. They didn't. Iowa's offense was somewhere in the sixties in early January, and with Bohannon's foot healing and Luka Garza breaking out, Iowa's offense has significantly improved. The defense has also bottomed out during that same time, which you would kind of expect to even things out, but maybe not.

Anyway, if we move back to our fantasy world with an improved defense and look at wins, we see that an improvement of 25 ranking spots (or approximately one fewer adjusted point allowed per 100 possessions) would be expected to yield about an extra half-a-conference win. Or, in other words, for every extra point per one hundred possessions that Iowa surrenders, they lose something like a half-a-game over the course of a conference schedule.

In terms of win probability, the relationship for every extra point per hundred possessions is three percentage points. For example, if Iowa has a 41% chance of beating Minnesota up north with the 250th ranked defense, according to this model (again, this matches pretty close to the 40% that Kenpom gives them), that goes up to 44% with the 225th ranked defense, and 47% at #200. That may not seem like a lot, but just having even an averagish defense (ranked somewhere near 175th) would give this team a nine percentage point bump. If you add in the bump a team gets from home court advantage, you are looking at somewhere near a 15 percentage point swing. That is a lot. 

Last year, Iowa's defense -- which would be wonderful right about now -- finished the season ranked 123rd in the country. A team with this year's offense and a defense that ranked 123rd would be expected win about eight or nine Big Ten games and add 15 percentage points of win probability per game compared to this year's iteration. That's not the top of the Big Ten by any means, but it's the difference between being a potential bubble team if things break right like last season and a cellar dweller like this season. 

All this is really just a fancy way of saying that defense matters, and that it is hard to overcome having a defense this bad even with an offense ranked in the Top 30 in the country. I certainly have more faith in Fran to build a top-notch offense than I do a suffocating defense, but I'm not willing say he flat out can't coach defense. Five of his eight Iowa squads have ranked 77th or higher in adjusted defensive efficiency, including three teams that ranked 34th or higher. Last year certainly wasn't great, and the fact that this is now his second Iowa team to rank worse than 200th on defense is nothing to celebrate. I'm optimistic that this group of players can learn to play as a more cohesive unit on defense. But how much better will they get? 

That really is the question. Experience should obviously bring improvement over the next few years, but there's a good chance that improvement could be limited if the roster isn't reconstructed with an influx of guards who can defend the perimeter. Next year brings in a potential offensive star in Joe Wieskamp and a relatively unknown shooter in C.J. Frederick. We should also get to see a full season from Connor McCaffery. (Knock on wood.) But it's probably not smart to expect those three to turn around Iowa's defense right away on their own. So the defensive improvement for next year likely comes down to the core of this team improving significantly at playing smarter team defense. The addition of a graduate transfer guard who can defend the perimeter and handle backup point guard duties would also help. It doesn't need to be another big time scorer, but if Fran can find an Anthony Clemmons-type of player out there, that would be a boon for this roster. If that doesn't happen, the future of this defense likely comes down to how much better guys like Jordan Bohannon, Isaiah Moss, Maishe Dailey, and Jack Nunge can get at defending the perimeter or how much better this team can get at implementing a zone. 

If significant improvement is made, then Iowa should be competing for Big Ten titles again in the near future. If only marginal improvement is made, we could be looking at something like the late-era Tom Crean Hoosiers teams who could fill it up on offense, but couldn't defend to save their lives. I'm hoping for the former, but if we get the latter, I just hope we also start getting the NCAA Tournament appearances that some of those Indiana teams had. 

9 Comments
View 9 Comments