Just 11 months ago, Kentucky basketball seemed nearly invincible. Coming off of a national championship, two straight Final Fours, and three consecutive trips to the regional finals -- each team built with a different core of superb freshman -- John Calipari appeared to have built an unstoppable machine. The machine was fueled by near NBA-ready talent, and the mechanism quickly prepared that talent to play aggressive, cohesive, and team-oriented ball. Coming into this season, with yet another top recruiting class led by center Nerlens Noel, most observers expected more of the same.
It didn't happen. As we sit today, at the beginning of March, no credible list of NCAA championship favorites includes the guys from Lexington. This version of the Wildcats simply isn't as good as the last. This is obvious; last season's team finished undefeated in the SEC, dominating the competition, while this season UK already has four conference losses with three games left to play.
Kentucky is still good, but they aren't great. Digging deeply into the differences between this season's team and the 2012 championship squad paints a picture as to why this is, and also highlights a few of the things that made Anthony Davis and company so great.
This year, Kentucky is worse than last season at both ends of the floor. A year ago, the Wildcats scored a blistering 1.20 points per possession during the conference season, while this year they have only managed 1.08 points per possession during SEC play. While these points per possession numbers are nice, it is possible to break things down further. Using a bit of mathematics, we can estimate how differences in a number of different statistical measures created these season-to-season changes. (The procedure used is described at the end of this article.)
To help visualize the year-to-year change in the performance of the UK offense and UK defense, I have presented how differences in various factors contribute using something called a waterfall chart. For those that spend their days in the business world, or simply are fans of the slide decks that commonly accompany corporate earnings reports, the waterfall charts will look familiar. The chart below shows the transformation of the Kentucky offense from 2012 to 2013.
Kentucky's offensive transformation
The bar all the way to the left shows the points per possession scored by Kentucky offense during the 2012 SEC regular season. The bar all the way to the right shows the same statistic for the 2013 SEC season. The bars in between these two bars show how year to year differences in various statistical categories have changed the points per possession totals. Green bars correspond to changes that have increased scoring efficiency, and red bars correspond to changes that have reduced scoring.
The first thing that jumps out in that chart is just how much of a difference turnovers are making for the Wildcats this season. Roughly half of the reduction in scoring efficiency for the Wildcats is due to an increase in turnovers. Turnovers are measured using turnover rate (TO%), which is simply the percentage of Kentucky possessions that end in a turnover. Last season, Calipari's team was fantastic at avoiding turnovers, only coughing up the ball in 15 percent of their possessions during SEC play. This year, they are below average; UK's 21 percent turnover percentage ranks ninth in the SEC.
The curious thing is just where this change in turnovers is coming from. Last season Kentucky point guard Marquis Teague was frankly a little shaky, turning the ball over in roughly 24 percent of the possessions that ended with the ball in his hands. It is usually a bad sign when a team's point guard has such a high turnover rate, but in Kentucky's case it didn't matter, because virtually everyone else who contributed significantly on offense seldom turned the ball over. Anthony Davis, Terrence Jones, Doron Lamb, and Darius Miller all combined to use a large fraction of UK possessions while rarely losing the ball. This season, most of the guys on the roster have turnover rates that are more typical of college basketball players.
The rest of the change in the Wildcat offense comes from a reduction in shooting percentage from three point range and at the free throw line. Coach Cal's team didn't take a lot of threes last year, but when they did they drilled them, hitting 41 percent from long distance during the conference season. This year UK has hit 36 percent from deep -- still a very respectable shooting percentage. The free throw shooting also contributes. A year ago UK was strong from the free throw line, making 74 percent during the conference season. Only one regular player (Terrence Jones) hit less than 70 percent from the line, while several Wildcats shot better than 80 percent on free throws. This season, Kentucky is hitting 65 percent of their free throws during SEC games, with five rotation players making fewer than 70 percent.
This season's defense has also taken a step back. The figure below shows the waterfall chart for the changes in the Kentucky defense from 2012 to 2013. No single category is the sole reason for the decline in the performance of the Wildcat D, but rather the change is due to the cumulative effect of a number of small factors. UK's defensive regression has been death by many tiny cuts.
Kentucky's defensive transformation
The single largest source of decline for the UK defense is in opponents' two point field goal percentage. Last season conference opponents only connected on 41 percent of their twos against the Wildcats, while this season they hit 44 percent of these shots. Using the season-long play-by-play data available at hoop-math, which includes non-conference games, this year opponents have connected on 56 percent of their attempts at the rim (compared with 53 percent last year) and 30 percent of their two point jump shots (compared with 26 percent). Kentucky doesn't block quite as many shots as they did last season, and additionally opponents are shooting for a somewhat higher percentage on unblocked shots.
The rest of the decline in the Wildcat D can be traced to a change in several factors. While John Calipari's teams typically do not force many turnovers, preferring instead to focus on defending the rim while avoiding fouls -- working to contain rather than disrupt -- this season UK opponents rarely turn the ball over. During conference play, Kentucky opponents are only turning the ball over in 15 percent of their possessions; this is the same low turnover rate that the great 2012 Wildcats managed to hold to while on offense. In other words, Kentucky's defense is putting so little pressure on teams that their average opponent protects the ball like a national championship team.
Kentucky has also regressed somewhat on the defensive glass (determined by opponent offensive rebounding rate, or ORB%), as well as in their tendency to put opponents on the free throw line (as measured by the change in free throw rate, or FTR). And Wildcat opponents have made more of their free throws this season, a factor that is at least partly due to bad luck.
The difference between being a good team and a great one may seem dramatic, but it often is not. It is sometimes nothing more than the accumulation of a bunch of little things that when taken together become something large; the case of the 2013 Kentucky Wildcats make this clear.
Explanation of the mathematics. Feel free to skip this. I have included it because some readers will care.
With a bit of algebra, it is possible to derive an equation that estimates points per possession based on the various percentages used in the two waterfall charts above. The details of that derivation are here, but a few additional comments are needed.
1. The derivation I linked makes one assumption -- that no offensive rebounds come on missed free throw attempts -- which is not quite right. Yet this assumption is almost right, and very helpful; it can be mostly corrected for with the addition of an empirical parameter.
Another consequence of this assumption is that it results in a somewhat different and idiosyncratic definition for offensive rebounding rate. To convert this strange rebounding rate (OR) used in the formula to the more common one (ORB%), I have found that the following empirical relation mostly works.
OR = ORB% + 0.03
That leaves us with the formula for points per possession
You can use this equation to amaze and impress your friends, but just remember that it is an approximation that is sometimes off by 0.01 or 0.02 points per possession.
2. I used the following approach to estimate the magnitude of the year-to-year change in each statistical category on points per possession.
First, I calculated the points per possession from the formula using this season's statistics, the points per possession if only one category was changed to last season's value, and the difference of these two numbers. For example, to calculate the effect of two point shooting percentage, I estimated points per possession for Kentucky with this season's statistics in every category except two point shooting percentage, which I set to the value from last season. This process gave a raw points per possession change caused by the change each statistical category.
Then, I added all of these effects up and compared this total with the actual season-to-season difference in points per possession. My estimated difference and the real difference were not the same, although they were close. In the case of the offense, my estimation of the difference was off by less than 0.004 points per possession, while in the case of the defense the estimate was off by 0.015 points per possession. To make the waterfall chart work, I rescaled each effect so the sum of the effect differences was equal to the actual points per possession difference. Only a small rescaling was required.