Skip to content

Kentucky Sports Radio

University of Kentucky Basketball, Football, and Recruiting news brought to you in the most ridiculous manner possible.

KenPom Bomb: Advanced Historical Trends

Given the product that was placed on the football field this fall one would be safe in assuming that Kentucky fans are relieved that basketball season is right around the corner.  If you remember all the way back to Tuesday morning, Ken Pomeroy released his annual preseason ratings that rank every single solitary college basketball team in their respective spot according to a very detailed algorithm (yes even the teams like #347 Grambling State are ranked here).  To someone who doesn’t study these rankings religiously, the valuable information on his site can become overwhelming and be ignored because of it.  But fear not, it’s not as complicated as it seems. So in order to give some clarification on the matter, I’ll explain some of the most important aspects of the ratings and just how they’re calculated.  In addition I’ll give some historical trinkets on teams who advance far in the tournament alongside some advanced John Calipari trends.

 

If you pay the $20 yearly subscription fee (and it’s highly recommended that you do), you’ll see on the individual team pages that there are 4 offensive and defensive statistics called the “Four Factors.”  These numbers began with basketball statistician, Dean Oliver.  These statistics are important because through much research, Oliver found that they’re the four highest correlating events in terms of successful basketball.  Effective Field Goal Percentage is the most important of the four and is just like its regular field goal percentage counterpart, except with 50% more credit given to three point shots.  Turnover percentage is the second highest correlating event; it’s simply turnovers divided by possessions.  It’s a tempo-free stat meaning teams like Wisconsin and North Carolina who play very different styles can be put fairly on the same scale without pace of play skewing the data.  Offensive rebounding percentage is third on the list; it’s simply percentage of offensive rebounds obtained in a tempo free formula (sensing a trend here?).  Last but not least there’s Free throw percentage, which is free throws attempted divided by total shots taken.

While these four numbers tell a great story of how each team plays, they’re not actually used in the calculations, they’re merely just a story of why a team is successful or not (I’ve been told this in conversations with Pomeroy).  There are three main numbers used in the calculations; points scored/allowed per possession, opponent points scored/allowed per possession, and D-1 average points per possession.  Points per possession are simply points scored divided by possessions. In a system that measures scoring margin like Pomeroy’s this is very important as fast tempo teams could be unfairly favored.  Here’s an example of a hypothetical calculation.  Last season Kentucky played a home game vs. Alabama and put in a gritty performance, winning 77-71 on 65 possessions.  While the raw numbers say this wasn’t an elite offensive performance that’s actually far from the truth once strength of defense is accounted for.  Alabama’s defense allowed 91.7 points per 100 possessions last season (raw), much better than the D-1 average so after adjusting Kentucky, theoretically, would have scored 82 points that day against average competition.  Five points doesn’t seem significant, but keep in mind after adjusting it would place them in an elite offensive category rather than above average.  It’s really as simple as that; if you play a team with a defense/offense better than average the raw rating will look better once adjusted and vice-versa for a bad offense/defense.  Every game is averaged and that’s how an individual team is rated.  Tons of math to get a simple, yet accurate result.

 

Now that the more complex calculations are out of the way here are some championship trends.

-The most well known trend is that the #1 or #2 ranked team in Adjusted Offensive Efficiency has won the national title 7 of the past 8 seasons.  Something significant to know; all teams have had very good defenses to pair with that offense as well.  In the season that didn’t see the #1 or #2 ranked offense win the title (2011), the two teams that knocked out the two most efficient offenses (Ohio State & Wisconsin) made the Final Four (Kentucky & Butler).  Indiana and Kentucky are predicted to be #1 and #2 respectively this year, both have good defenses as well.

-Rick Pitino’s best teams pair elite offense with elite defense (1996 UK and 2005 UofL), this season’s team is predicted to have the 34th best offense.  The Cardinals return the 103rd most efficient offense that lost its two most efficient players in Kyle Kuric and Chris Smith.  It’s hard to envision a way they can fit the championship trend.

-Last season when media elite were picking North Carolina to cut the nets down in New Orleans, Pomeroy’s system correctly predicted that Kentucky was the nation’s most elite team in the preseason.

-In terms of the Pythagorean Rating System (which Pomeroy uses), this is predicted to be the weakest year in the efficiency era (03 to present).  Only 9 teams are predicted to have ratings that exceed .9000 (higher the better).  Last season 18 teams finished with ratings above .9000.  17 teams did in 2011, 23 in 2010, 20 in 2009, and 31 in 2008 (Note that his preseason ratings are a tad conservative).

-John Calipari’s most elite teams have two commonalities that are often overlooked by basic stats, but not advanced stats.  His most successful teams have low turnover rates and block a high percentage of opponent shots.  His most recent Final Four teams have all been top 21 nationally in turnover percentage (each 17.0% or under).  Cal’s last three teams to make the Four have been ranked 1st, 6th, and 11th in block percentage respectively.  This season Kentucky has two elite shot blockers in Willie Cauley-Stein and Nerlens Noel to pair with a Point Guard in Ryan Harrow who is not prone to turnovers.

Article written by Jonathan Schuette

12 Comments for KenPom Bomb: Advanced Historical Trends



  1. Wow
    10:33 am November 1, 2012 Permalink

    Yawn



  2. Twitter Account Denied
    10:36 am November 1, 2012 Permalink

    So you’re telling me there is a chance. I read ya.



  3. Chicago Chris
    10:46 am November 1, 2012 Permalink

    “Last season when media elite were picking North Carolina to cut the nets down…”

    Huh?



  4. Jordan
    10:47 am November 1, 2012 Permalink

    Pretty much every analyst I’ve read says UK is extremely overrated at #3 right now. I tend to agree, probably should be more in the 10-15 range. But by March, we could be back in the top 5 and capable of another championship. To put the #3 ranking on these freshmen is unfair, though.



  5. capt
    10:48 am November 1, 2012 Permalink

    I like it.

    3) yes, that’s the way I remember it, at least early on.



  6. Chicago Chris
    10:48 am November 1, 2012 Permalink

    Thank you for the informative article. One comment: “elite” is so overused, it loses its meaning here. But a good article nonetheless.



  7. Economonster
    11:01 am November 1, 2012 Permalink

    Great write-up, I can read about basketball metrics all day. I wonder, if you pay the $20 sub fee to Pomeroy, can you download his data sets?



  8. me
    12:16 pm November 1, 2012 Permalink

    I still don’t understand how he puts this together before anybody plays a game. What in the world are UK’s stats based on? If it’s complicated math using predictions as inputs, then you’re no better off than looking at the AP poll. What am I missing?



  9. Jonathan Schuette
    12:25 pm November 1, 2012 Permalink

    #7. Yes you can. There’s a link that says “data” on top of multiple pages including the main one that’s free and available to the non-paying public. While the formulas are obviously not there you can still download the data in convenient excel form and play with it any way you choose from there.

    #8. I understand the skepticism, it will be far from 100% correct, but it’s usually pretty good. He tracks how typical players progress through their careers, how coaches use those players, and even know uses freshman offensive ratings. So much detailed thought goes into this. That being said it’s impossible to predict anything 100%.



  10. John Wall
    12:36 pm November 1, 2012 Permalink

    So basically louisville isnt going to beat anybody haha



  11. James
    12:36 pm November 1, 2012 Permalink

    Jonathan Schuette you suck. Please stop posting articles.



  12. Syrin
    4:51 pm November 1, 2012 Permalink

    This is great info, but where does Ken Pomeroy get his preseason figures from for a team like UK who will start 5 players that have put up no college numbers to analyze. is it purely his best guess as to the numbers they will post?

    11. You’re an ass. Please stop posting. We are all dumber for having seen your post.