Examining how pace affects the KU basketball team's defense


Ready to see something strange?

OK, we're going to have to get a little bit nerdy, but stick with me, because this is interesting stuff.

Here is a graph I charted that shows the relationship between the pace of the Kansas basketball team's games (number of possessions) and KU's defensive efficiency (number of points given up per possession).

Notice the downward trend? We'll get a little more into what this might mean later.

The confidence level of this data is 99 percent*, meaning there is only a one-percent chance that the trend of this graph is due to chance.

* — In many studies, a 95-percent confidence level is needed to prove the data is correlated.

What's interesting is that this kind of a correlation doesn't happen often. I checked out the Game Plan section for each Big 12 team on, and no other conference team had a correlation between pace and offensive or defensive efficiency with a confidence level of 95 percent or better.

Yet, KU's correlation between pace and defensive efficiency had a confidence level of 99 percent.

Just for good measure, I wanted to test to make sure our numbers were accurate.

Webprince from the message boards (You can read his KU basketball-based statistical analysis on his blog, Sports and Numbers) provided me with the adjusted defensive efficiency numbers from each game. This rewards KU for playing better defense against tougher offensive teams, and penalizes it for playing weaker defense against poor offensive teams. It also takes into account where a game is played (road, home, etc.).

The graph looks nearly the same.

The confidence level of this data is 98.4 percent. Again, we have a correlation.

OK, let's take a step back and examine what this means. It could be one of two things.

1. Slower tempo is causing KU's opponents to score more points per possession.

Or ...

2. Correlation does not equal causation, meaning that something else is causing the better opponents' efficiency that happens to coincide with slower tempo.

But what could it be?

Here are two of my own guesses:

1. Though KU is a good defensive team, it does not force many turnovers. The statistics back this up, as KU's opponents turn it over on 20.7 percent of their possessions (173rd nationally).

Because of KU's inability to get those turnovers, teams that are patient don't have a great risk of turning it over, and therefore have a greater likelihood of getting a better shot.

Webprince provided me a statistic that might back this claim up.

Through running his numbers, he found that KU's pace and opponents' turnover percentage was positively correlated with a 95.9 percent significance.

In English? As the pace goes up, KU's opponents' turnover percentage goes up. But as the pace goes down, KU's opponents' turnover percentage goes down.

This would seem counter to what I would guess. You would think fewer possessions in a game would mean that the possessions would be longer. And the longer the possessions are, the more likely a team would be to turn it over against KU.

The numbers suggest the opposite. The fewer the possessions in a game, the less likely a team is to turn the ball over against KU.

Here's a second guess to explain our data above:

2. KU's players don't defend well on long defensive possessions. Hypothetically, this could be because of fatigue.

KU has a relatively short bench at this time (31.2 percent of its minutes come from the bench, which is 183rd nationally). So is it possible that KU's players could be tiring after having to defend for 25-35 seconds?*

* — I know a lot of folks weren't happy with Colorado coach Jeff Bzdelik's comments after last game when he said: "Well, Kansas doesn’t want to guard our stuff. They hate that. So, you can’t accommodate them by taking quick shots and letting them off the hook. You need to make them defend.”

He's looking smarter toward the end of this blog, isn't he?

It's interesting to think that KU might want to try to speed teams up not because it will play better offensively, but instead, because it will play better defensively.

Any other theories out there to explain the numbers above?


fan4ever 12 years, 3 months ago

Excellent analysis. The abilities of Chalmers, Robinson and team to make steals a couple of years ago was a huge defensive advantage. Some of the guys on this team appear to have the same quickness, but just not the same mindset. At crunch time, it is those steals that drive all teams crazy and wins games.

KUAdam 12 years, 3 months ago

How many of those high-possession games were blowouts? If there were a big score separation between the two teams, the opponent tries to play catch-up, and takes quicker shots and leads to more possessions.

In a tight game, teams use more of the shot-clock trying to set-up a specific play. The game is tight, they want good shots only.

In a blowout, the opponent is just trying to get the first available decent shot in the hopes of making some to get back in the game.

I have no idea if this is the case, but it would make some sense. I am sure there are outlier games, but in general I wonder if this is the case.

Jesse Newell 12 years, 3 months ago

KUAdam — I like the thought. If that was the case, though, we might expect other Big 12 teams to show the same tendencies as KU. Only KU shows a correlation between pace and defensive efficiency with a confidence level of above 95 percent in the conference. And KU's is at a 99-percent confidence level.

But, then again, KU is different from a lot of other Big 12 schools in that it has played in more blowouts. That might affect the numbers a bit.

Remember, KU is No. 1 in the nation in scoring margin. I went ahead and looked at the other top 10 teams nationally in scoring margin to see if their numbers were like KU's. Turns out that one other team in the top 10 in scoring margin shows a (negative) correlation between pace and defensive efficiency — and that is Syracuse (with a 95-percent confidence level).

So perhaps your "blowout" theory could affect some teams' numbers.

shafs 12 years, 3 months ago

Your analysis is flawed. The data you show do not have a 99% confidence limit. In fact, plotting the raw defensive efficiency data you show results in a correlation coefficient of only 0.3, meaning that IF there is a correlation, it's very weak. Whatever correlation there appears to be, it's only because of the one high possession game where we had a defensive efficiency under 0.4. If this game is considered an outlier (which it statistically is), there is almost no correlation amongst the data.

That being said, your rationale about KU not forcing turnovers may be correct, but it's not shown by these data.

5a5quatch 12 years, 3 months ago

I would suggest you consider two things: 1) there may be a fundamental flaw in your theory because more turnovers are the CAUSE of higher pace, and, naturally, higher def eff (if so, one might have expected other teams' numbers to reflect the same thing...however, KU is so dismal at causing turnovers you probably should compare KU's graph to other teams with low TO rates...if KU's graph still stands alone, I still think this flaw is worth considering and KU's graph shows the CAUSAL correlation more strongly than anyone else because fast pace does not NEGATIVELY affect their defense, due to depth and talent, and other teams with low TO rates may not like to get sped up...KU plays great defense either way, and, when they get TOs, their numbers spike) 2) home court may explain some of this (does KU get more TOs at home than away? does pace speed up at home vs away? you should do separate graphs for home and away, and see if def eff behaves the same way. it could be that KU is better able to force TOs at home, or simply turn up the pace at home, and the numbers simply reflect a much better than average home court advantage).

Eliott Reeder 12 years, 3 months ago

I am a math moron, but this stuff still fascinates me. As far as KU's tendencies from the eye of pure observation, Bzdelik's comments certainly didn't shock me. It was that he had the nerve to come out and say it to the media that shocked me. Up to now it has been pretty obvious that KU doesn't like to play against teams with deliberate, slow-paced offenses. As pace quickens, the play becomes inherently sloppier, which leads to more turnovers. Isn't that the case for every team? Your thought that the opposite would occur (long possessions = greater opportunity to cause a turnover) is just not feasible. Slow pace = deliberate passing, well-set screens, finding the open man, good shot selection. Fast pace = riskier passing, sloppier execution, wild drives to the basket, rushed shots, etc. The numbers are interesting, but I think this is obvious at the fundamental level of basketball theory. Comparing this team to the '08 team, which so many love to do, is not really fair. We will probably NEVER have a combination of guards like RR and Mario, who worked so incredibly well in tandem that it was hard to tell who was truly responsible for each steal. The current team is just starting to develop its defensive chops. People seem to forget how young we are outside of Sherron and Cole (who still kinda plays young). I think our current guards, esp. Tyshawn and Elijah, will be defensive giants in a couple of years.

Brandon Deines 12 years, 3 months ago

I'm only guessing, but I would wager that our offensive efficiency chart would look very similar (except that the line would tilt the other way, of course).

lance1jhawk 12 years, 3 months ago

Whoa.... did I just learn something.... Nobody told me there was going to be charts and graphs.... I'm going to go take a nap and take all this in.

All kidding aside, nice work Jesse

Jesse Newell 12 years, 3 months ago

5a5quatch — I've added additional graphs based on your comments:

5a5quatch — There does appear to be more of a correlation at home than on the road (I included the La Salle game in KC as a home game).

Mrs_Estherhouse — The offensive efficiency chart doesn't look the same as the defensive one, as you can see.

pizzashuttle 12 years, 3 months ago

I'd attribute some of the correlation to how young the team is. When Tyshawn, X and Marcus are in the game together that's 2 soph's and a freshman. IMO younger players get bored defending a slower paced, half court offense whereas a team with older players has more mental toughness and can focus on defense for a longer period of time. We had more experience on the '07-08 team which was the reason they could fight their way back into games when they would get down big. When the other team slowed the pace to milk the clock, their experience allowed them to focus more on defense which resulted in more turnovers during longer possessions.

FairgroveJayhawk 12 years, 3 months ago

To go along with your first assessment of KU being a good defensive team but not causing turnovers as the 08 team did:

This team seems to get more turnovers by stepping into the passing lanes instead of picking pockets. Teams running faster are more likely to turn the ball over more while passing. The cupcakes early in the season could be effecting the data as they provided many easy baskets. Cupcakes are also likely to play into a faster tempo even if they are not typically a 'fast' team that would also allow more opponent turnovers.

It would be interesting to consider running these numbers between conference and non conference schedules to remove the cupcake factor.... but then again we will likely improve as a team defensively as the season progresses.

No, Jeff Bzdelik's was just being a sore loser. If he was smart enough to figure out such correlations, he would win more games.

Mike Kendall 12 years, 3 months ago

Well done, Jesse! Great read, as always!

Mike Kendall 12 years, 3 months ago


I have one thing to say in regard to your last question, "Will we have to rely on buzzer beaters thoughout the NCAA tournament?" If that is the case, I will be resting in peace, six feet under, having died of a heart attack!!!!!!!!

melrank 12 years, 3 months ago

Maybe this comment supports some of the theories and coefficients and correlations, but I graduated from the business school and hated econ, so I'm not sure what all of those fancy words mean.

To me, this is pretty simple: There are only a few teams who want to get into a track meet with us so the general strategy to beat Kansas would be to limit possessions and keep it close. If our opponent is playing well or we are playing poorly, then the strategy of keeping it close is working thus reinforcing the low possession strategy AND probably the poor defensive efficiency.

If we play well and hit shots, they play poor and don't hit shots, the possessions will most likely increase as we get a lead and they abandon their strategy in order to catch up thus reinforcing the opposite result of high possessions AND probably good defensive efficiency.

Turnovers obviously contribute to the end result.

The poor defense on longer possession theory is a stretch. Let's hit 5 or 6 of the bunnies we missed and shoot even 65% from the line against Colorado and Buzzlick's comments become stupid. Hell, they only shot 37% from the field at home. How bad was our defense?

Mike Kendall 12 years, 3 months ago


Defense was stinkin' bad!!!!!

melrank 12 years, 3 months ago

Mike - opinions can vary wildly when emotion kicks in, but the facts are pretty simple:

Colorado shot .376 from the field which is about 20% lower than their season average of .473.

From 3-point land, Colorado shot .296 which is about 23% lower than their season average of .375. AND they were playing at home.

Want more stats?

Nationally, KU is right at the top at 36.3% defensive FG made, so we held them at just 1/100 of our season average AT their place when we played a terrible offensive game contributing to the slower pace and their ability to prolong and stay in the game.

At .375, which is what Colorado shot, that would be good for 9th place nationally in FG percentage defense - 9th place in the nation!

Again, I will ask, how bad was our defense.

melrank 12 years, 3 months ago

I'm too lazy to figure out what our defensive FG % would be on the ROAD this year (or any year), but I'd bet it would be higher than the .376 that Colorado just rang up.

5a5quatch 12 years, 3 months ago

Jesse, thanks for the addl Home/Away graphs. Not surprisingly, it looks like opponenets are better able to control tempo (slow it down) when they are not in AFH. The interesting thing to me, however, is that KU's performance in def eff is virtually the same in the two settings as #poss goes up to ~75 (~.8), but KU actually handles the slow down BETTER on the road. As #poss approaches 65, KU's road def eff is around .86 (nearly flat wrt 75 poss), while its home def eff is around .95.

I would have expected that the diff in the Home/Away graphs would have been seen in better def eff as #poss goes up at home. In reality, the diff in the slope appears to be more due to KU NOT performing well in a slow down setting at home. They actually do it pretty well on the road.

Interesting info! A few explanations jump to mind: 1) most of our early season games were at home, and KU was not yet effective at defending throughout a long clock, which skewed the home statistics, 2) only good teams are able to slow things down at AFH, and they score at better rates than the bad teams, 3) KU expects this tactic on the road and focuses better.

These theories (and others) could be examined via more graphs (early home vs late home, good home opp vs bad home opp, etc), but at this point the data sets are getting small. My guess is something intuitively makes sense by looking at the home slow down opponents and dates.

Again, interesting analysis. As you can tell, I love this stuff.

melrank 12 years, 3 months ago

Jesse - do a study about why teams shoot lights out against us from 3.

We are one of the best in the Nation at total FG% defense, but somewhere in the 80's I believe at 3-pt FG% defense.

That is a question many fans would love to know the answer to.

Keep up this good stuff.

usafJayhawk 12 years, 3 months ago

You wrote an article and used the term "confidence intervals." that's bold!

Commenting has been disabled for this item.