Turnover Improvement: Real or Mirage?

[Promoted from the Diaries. -- Ed.]

I think it's fair to say that turnovers are on the mind of the Big Blue Faithful heading in to the annual tilt with Louisville.  We've seen improvement in the number of turnovers committed by the Cats in their last four games, but there is still some question as to how much of that improvement reflects real growth by the team and how much is a result of playing inferior teams.

There is some anecdotal evidence that players, particularly Michael Porter, are becoming more comfortable with their role on the floor and are making fewer mistakes leading to fewer turnovers.  Still, there's a nagging doubt (for me anyway) that the gains made over the past two weeks are a mirage of competition, and that on Sunday against a good UL team that plays the kind of pressure defense that has given UK trouble this season the same turnover bugaboos that plagued the squad agianst UNC, VMI, Kansas State, West Virginia, and Indiana will reappear.

With that question in mind, I decided to take a closer look at UK's performance and that of its competition to see if there was something in the numbers to backup what my eyes have told me and what the players and coaches have said.

The Basics

Here are UK's turnovers and possessions for all their games so far this season.  Stats come from

Opponent VMI UNC Del St Longwood KSU WVU Lamar Miami (FL) MVSU Indiana App St Tenn St Florida Atl Cent Mich
Possessions 92 73 65 82 76 64 77 73 72 74 78 77 64 68
UK Turnovers 25 28 12 19 31 23 16 14 14 23 15 16 7 13

In their first ten games UK committed 205 turnovers, 20.5 per game.
In their last four games UK committed 51 turnovers, 12.75 per game.

Of course, games are played at different paces and this can affect the number of turnovers committed.

In their first ten games UK had 748 possessions and turned the ball over 27.4% of the time.
In their last four games UK had 287 possessions and turned the ball over 17.8% of the time.

That's pretty good.  UK averages about the same number of possessions per game in each group (74.8 vs 71.7) but is committing roughly 35% fewer turnovers in its last four.  This tells us that UK's turnover numbers are not being artificially deflated by a slower pace over its last four games.  So far so good, the improvement looks real.

The Competition

The real meat and potatoes of this analysis though is to look at the teams UK has played and try to determine the extent to which the quality of the competition affects the number of turnovers UK commits.  The way I have chosen to do this is to get an Expected Number of Turnovers in each game based on the number of possessions and each opponents' defensive TO% (supplied by Ken Pomeroy and current as of Jan. 2).

Opponent VMI UNC Del St Longwood KSU WVU Lamar Miami (FL) MVSU Indiana App St Tenn St Florida Atl Cent Mich
Possessions 92 73 65 82 76 64 77 73 72 74 78 77 64 68
Opp def T0% 26.5 24.6 16.9 25.5 27.0 28.8 19.2 19.2 21.4 25.1 17.6 20.1 16.8 22.9
Exp. UK TO 24.4 18.0 11.0 20.9 20.5 18.4 14.8 14.0 15.4 18.6 13.7 15.5 10.8 15.6
Actual TO 25 28 12 19 31 23 16 14 14 23 15 16 7 13

There's some good news and bad news here.  The bad news is that over the last four games the turnovers are about what we would expect them to be given the number of possessions in the game.  This suggests that competition was a big factor in the apparent recent improvement.  The good news is that aside from the two games against UNC and Kansas St and to a lesser extent the games against West Virginia and Indiana, UK has performed 'as expected' against its competition.  This might imply that the team is actually closer to average than we thought when it comes to taking care of the ball.

On the downside, I don't think the results against UNC and Kansas St. can simply be dismissed as outliers.  Both of those teams play pressure defense and in each case UK turned the ball over far more than expected and more than any other team has that has faced UNC and KSU.  It could very well be the case that given its lack of a true or experienced PG UK is more susceptible to this kind of pressure defense than other teams are.  That certainly does not bode well for Sunday's game.

The Competition, Part II

There's another way to break down these numbers.  Turnovers aren't all the same.  Generally they can be divided into two categories: Dead ball Turnovers and Live ball Turnovers (aka Steals).  As the names imply, Dead ball TOs are turnovers that result in a stoppage in play (walks, five second calls, stepping out of bounds, etc).  Live ball TOs (henceforth referred to as Steals) do not involve a stoppage in play.

There are other differences.  Steals often lead to fast break opportunities whereas Dead ball TOs require the receiving team to inbound the ball and allow the defense to get set.  More importantly (for the purposes of this post), steals are usually caused by direct physical action by the defense whereas dead ball turnovers are usually caused by mental mistakes by the offense.  This is not to say the defense has nothing to do with forcing dead ball turnovers or that mental mistakes don't lead to steals, but speaking generally the defense has more control over forcing steals and the offense has more control over committing dead ball turnovers.

We can account for this in breaking down the game results for UK.  Instead of using the Opponents defensive TO% (as I did above) to get an expected TO count for each game, we can divide turnovers into the two groups and get an expected value for each.  This should allow us to more accurately assess the affects of competition.

Opponents Steal% is easy to get as it is supplied by Ken Pomeroy.  Kentucky's Dead ball TO% can be found by taking the Offensive TO% (how often the offense turns the ball over) and subtracting the Offensive Steal% (how often the offense gets the ball stolen from it).  These numbers are, again, supplied by Pomeroy.  As a quick aside, both percentages have a common denominator (possessions) so we are mathematically justified in subtracting one fraction from the other.

Opponent VMI UNC Del St Longwood KSU WVU Lamar Miami (FL) MVSU Indiana App St Tenn St Florida Atl Cent Mich
Possessions 92 73 65 82 76 64 77 73 72 74 78 77 64 68
Opp Steal% 15.8 13.7 7.8 10.9 10.3 12.9 9.5 9.3 10.3 11.7 8.1 10.3 6.4 11.7
Ex Opp. Steals 14.5 10.0 5.1 8.9 7.8 8.3 7.3 6.8 7.4 8.7 6.3 7.9 4.1 8.0
Actual Steals 13 16 5 5 10 6 9 6 7 9 8 8 2 7
UK dbTO% 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6 13.6
Ex db TOs 12.5 9.9 8.8 11.2 10.3 8.7 10.5 9.9 9.8 10.1 10.6 10.5 8.7 9.2
Actual db TOs 12 12 7 14 21 17 7 8 7 14 7 8 5 6

There's a lot of numbers up there so here's the summary:

Exp. Opp. Steals Actual Steals Exp. UK db TOs Actual UK db TOs
First 10 Games 84.8 86 101.7 119
Last 4 Games 26.3 25 39 26

Looked at this way, there are some reasons for major optimism.  Throughout all 14 games UK has had the ball stolen pretty much exactly the number of times we would expect with practically no difference between the first ten games and the last four.  Contrary to what the previous results showed, this suggests that UK might not have any weakness to a particular type of defense after all. 

Where the team has made great strides is in cutting down on the number of dead ball turnovers.  That makes a lot of sense if you think about it.  As I mentioned earlier, deadball turnovers are mostly due to mental mistakes.  As players become more comfortable with their roles and with each other on the court, those kinds of mistakes should decrease.  The improvement probably isn't quite as good as what is shown because the defense will have some indirect affect on those turnovers, but I feel comfortable in saying that improvement is real and not due to competition.

So You're Saying There's A Chance?

What does all this say about the team heading into Sunday's game against Louisville?  The Cards steal the ball 12.4% of the time, good for 39th in the nation.  I think we should expect that to continue on Sunday and maybe go a bit higher since they are at home.  UL plays at the same tempo as UK (71.4 adj pos/40 min versus 71.3).  I would put the over/under on total turnovers at 18.  If UK commits fewer than that they will be in the game with a good chance to win.  If UK commits more than 18 turnovers, it will be a long afternoon.