Finding another "elite" stat is a key for UofL
Louisville is without a doubt one of the elite teams in college basketball this season. They have taken up near-permanent residence in the top 10 in both the AP Poll and KenPom rankings. They were a 2 seed in the NCAA's first release of the rankings. However, UofL is a team that is very good to great in a number of areas, but elite in comparatively few. Today, I look at several key stat categories to determine where improvement would have the most impact for UofL. Because this article is a bit more technical and analytics-heavy than others, I'm going to start off by summarizing the key takeaway.
Takeaway: Limiting opponent 3 point shooting is the stat that can have the biggest impact for UofL, as they allow a relatively high number of 3 point shots. The perimeter trio of David Johnson, Ryan McMahon, and Samuell Williamson has shown an ability to defend the 3 pointer, and UofL should look to give them more time together.
Picking the Stat Categories
First, I'm going to pick stat categories that are easily measurable and have public data. I pulled the data from KenPom.com and hoop-math.com, so thanks to them for having a ton of data available for every team in the country. Here are the stats I picked:
% of shots at the rim
% of shots as 2pt jumpers
% of shots as threes
Free throws per field goal attempt
FG% at the rim
FG% on 2pt jumpers
FG% from three
Free throw percentage
Defensive rebound rate
Offensive rebound rate
Turnover rate
Block rate
All of the above, but for opponents
That feels like a lot when I type it out like that. Hey, the more the merrier!
Defining Improvement
In another step to simplify things, I'm going to refer to Kentucky's rank in these categories by percentiles. For example, 95th percentile means that UofL is better than 95% of the country; that ranks them about 18th or so. I decided to look at the impact of improving Louisville's percentile ranking by 10; for example, going from 50th to 60th percentile.
One thing to consider is, because I'm using a 10 percentile improvement as the standard, anything where UofL ranks at the 90th percentile or better is not eligible to be listed. That means 3pt FG% and opponent FG% at the rim are out, because UofL is already elite at those.
Defining Impact
I decided to convert the improvement in a stat into points per game. For example, improvement in defensive rebounding reduces opponent's possessions, and I multiplied the reduction in possessions by the opponents' points per possession. I'm happy to share the formulas for everything if anyone wants it, but it's a lot.
Most stats are pretty straightforward, but there are a few exceptions. An increase in shots at the rim has to come at the expense of 2 points jumpers and threes, and an increase in free throws comes at the expense of field goal attempts.
In all cases, I left all of the other stats at UofL's current season performance when calculating the effect of improvement. So, what are the results?
Very Good to Elite is the key
The most impactful stat to improve is opponent 3 point FG%. Improving from 30% allowed to 28% would be worth 1.6 points per game for Louisville. This is because UofL allows a fairly high number of 3 point shots, ranking in the 43rd percentile nationally.
You'll notice that the top 4 most impactful stats are all things UofL is very good at. They are in the 85th-89th percentile in all of these. Inversely, UofL is bad at forcing opponents into 2 point jumpers, and improvement there would also be impactful. That's not a coincidence...
Improvement is most important at the extremes
The fact is that the biggest impact from improving a few spots in a stat ranking is always going to come at either extreme: the very best or very worst. It's because most team stats in college basketball tend to follow a specific distribution pattern where there's a big dropoff among the top or bottom handful of teams. Just look at an example:
The nation's leader in offensive rebound rate is West Virginia at 40.4%. The gap between them and 10th place Duke is 4.6%; that's the same as the gap between Duke and Kansas, who ranks 73rd. In percentile terms, moving about 3 percentiles from Duke to West Virginia would yield a similar improvement to move 18 percentile points from Kansas to Duke.
You'll see a similar distribution in other, unrelated stats. For example, here's opponent FG% at the rim:
With these stats (and most other team stats), you're going to see a similar pattern. Improving from being in the top 5% of something to the top 1% is a bigger deal than improving from the top 15% to the top 5%. The takeaway is that it's a huge advantage to be among the very best handful of teams in the country at something, and not be among the very worst.
What should Louisville do?
As I mentioned before, Louisville is elite in relatively few stats categories. 3pt FG% is the only stat where UofL ranks in the top 10 nationally per KenPom, and that has been the Cards' Achilles heel in losses to Texas Tech, Kentucky, and Georgia Tech. If UofL could go from very good to elite in other categories, it would help them have a calling card in their biggest matchups. By my analysis, defending opposing 3 point shooters would be the most impactful area to take from very good to elite. But how can they do that?
A key appears to be more length and athleticism on the perimeter, with strong communication. Since David Johnson began getting regular playing time, UofL has been much more effective at limiting opponents from deep when he, Ryan McMahon, and Samuell Williamson play together. Since UofL's win streak began Jan 7th, opponents have shot just 24% from deep when this trio was defending the perimeter. But opponents are hitting 32% against any other perimeter combination. The last game in Atlanta against Georgia Tech is a great example. This trio held the Yellow Jackets to 0-4 from deep, while Georgia Tech went 4-10 against other lineups. Watching the film, a few key factors stand out. First, Johnson has become very good holding up at the point of attack against dribblers and keeping them out of the lane. Second, Williamson is able to help on drives and has gotten better at recovering to contest jump shots. Lastly, McMahon does a good job of being alert on the perimeter and organizing perimeter rotations. Georgia Tech didn't get quality looks against this trio; instead, every perimeter was either off of a desperate kickout pass when a drive failed, or was the result of the shot clock winding down and no good options materializing. That's the recipe to control an opponent's three point shooting, and this trio in particular is doing a good job.
Chris Mack has indicated a willingness to shake up the starting lineup and rotations for UofL. I think it's absolutely worth a shot to give this trio extended time together, as I think their play can be a key in defending 3 point shooting. If UofL can become elite at that, they will have another major weapon as they look to contend for a national championship.
Thanks for reading my newsletter. For a primer on the stats I like to use, click here: http://hoopsinsight.com/stats-glossary. If you'd like to take a look at previous newsletters I've sent, check out my archive. If you have any questions, want to argue a point, or have some feedback, feel free to reach out via email at sean@hoopsinsight.com, or on Twitter @hoopsinsights. If you liked this, let me know as well, and tell your friends to subscribe at www.hoopsinsight.com.
You can forward this to others, but please ask them to subscribe as well so I can keep track of who's enjoying my insight and analysis.