Similarity Previews: KC-Area Teams

[NOTE: Some of this is cross posted over at Upon Further Review, a great KC-area sports blog.]

With the success of the similarity predictions so far (WARNING: extremely small sample size):



I figured I’d continue down this path, at least until I start to see some bad results.  I’ll be doing a comprehensive preview of the first round games that highlights the instances where the similarity prediction dovetails from the standard Pomeroy prediction, but first I’m going to tackle the match ups involving teams that are of interest to UFR readers.  Any system that pegs the KU and KSU games as anything other than huge mismatches probably involves throwing things off the Empire State Building, so I won’t spend much time on those games.  The more interesting game is Missouri-Clemson, so that’s the one I’ll start with, and focus on.

MISSOURI vs. CLEMSON

One of the most fascinating aspects of this game is the fact that these defenses are mirror images.  Take a look at the 2010 defensive similarity tables for each team:



 

[You may notice that I’ve changed things a bit from my initial UFR posts.  The concept is still the same, but now I split the similarity scores up so that the offense and the defense get separate comps.  I think this better reflects reality, and allows for more accurate comparisons, since two teams may be twins on one side of the ball, but very different on the other.  I’ve also added a listing of which categories had the highest weight in the similarity score calculation, so you can see why the teams are considered similar.  And lastly, I am limiting comps to major conference teams, plus a few others like Xavier, Gonzaga, etc.  Because being similar to, for example, 2005 Murray State doesn’t really mean much to me.]

Clemson and MU’s are each other’s top defensive comps, due to their efficiency, steals, turnovers, and eFG% defense.  Clemson is slightly more balanced, in that Missouri relies a little more on those turnovers to make up for sub-par rebounding, but other than that we’re looking at the same defenses.  I was going to say that this means the offenses should be well-prepared, since they see these mirror images every day in practice, until I remembered this isn’t football – the same players are on offense and defense, and they can’t guard themselves in practice.  The caveat is that Missouri’s Basketball Socialism (no player averages more than 27 minutes per game) may mean that their bench players are pretty good stand-ins for the starters, so their offense may be more used to practicing against this type of pressure.  Let’s take a look at the comps for those offenses.  I’m including teams from past seasons this time:



Four out of these five past comps were held under a point per possession in their first postseason game against a good defense (which Missouri most certainly is), and the exception (2007 Purdue) only managed 1.03 PPP, which is no great shakes.  I’m not sure why these teams would struggle, as they can shoot and rebound reasonably well, and don’t have serious turnover problems.  I think the main thing is just that they are simply not great offenses, and a good defense will have the upper hand against an only-above-average offense.



The postseason performance of these comps is slightly better, but there’s not a huge difference.  I think the key here may be that Missouri is quite a bit better at holding onto the ball than Clemson is, which may mean they end up with an advantage in FGA, always important in a rough, defense-oriented game.

As I mentioned at the top, I’ve developed a system for making game predictions based on the similarity scores.  The basic concept is that I look at how a team has performed this year against squads that are similar to their upcoming opponent, and make a prediction based on that.  For more detail, see my introductory post.  Here are the relevant numbers for the (Yay!) Tigers:



As usual, green equals good, and red equals bad.  On the left side above, you can see that when Mizzou played opponents whose defenses are most similar to Clemson, their offense performed basically at their season average efficiency. (This is not always the case, as we’ll see below).  Ditto for the MU defense in games against offenses that are similar to Clemson.  Now let’s look at the (Boo!) Tigers:



The left side shows first of all that Clemson hasn’t faced any defenses very similar to Missouri (the highest SIM is only 75); and second, that when we look at games against opponents that were kind of similar, the Tigers have on average been a couple of points better than usual.  However, on the right side we can see that when facing offenses that resemble MU’s, Clemson’s defense has been slightly worse than usual (lower is better for defense).  Here is what we get when we put it all together:



The left column shows the standard Pomeroy prediction for the game, and the right side shows the prediction based on the two tables above.  Pomeroy has Clemson as an ever-so-slight favorite, winning 52% of the time, with a margin-of-victory of 0.5 points.  When I attempt to account for opponent style, the odds shift a few percentage points over to Missouri’s side, but that 3% swing is essentially irrelevant.  Both systems are throwing their hands up in the air on this one.  The average sportsbook line as I write this is Clemson –1, so the gambling markets are similarly unsure.

KANSAS STATE vs. NORTH TEXAS

[EDITOR’S NOTE: I originally thought KSU’s opening round opponent was UC Santa Barbara, for some reason.  The charts and writings have been corrected to reflect the true opponent, North Texas.  To be honest, the article was more interesting when the opponent was UCSB, as the similarity and Pomeroy predictions varied by about 8 points.  Oh well…]
There’s no need to go into as much detail on this one, so I’ll just show you North Texas’s comps, and then get straight to the prediction.



The Mean Green have an NCAA average offense, so all these major conference comps are considerably better than North Texas.  But, you can see one of the things they do well is get to the free throw line.  Considering how often the Wildcats foul, they may be shooting a ton of free throws.  Not that it should matter in this game.



Their defense is even more average than their offense, though they do rebound decently.  Given that Dennis Clemente will probably try a shot like this one at some point during the game, there may be some misses for North Texas to clean up.  Especially since they won’t make KSU turn it over, giving them even more looks at the basket.  This style analysis is mostly irrelevant for this game, however – check out the prediction:



The standard Pomeroy prediction already favors KSU by 19, and the similarity prediction agrees, only adding a half a point.  As you’ll see below, that 95% is nearly a #1-seed-esque advantage.

KANSAS vs. LEHIGH

Same plan here – the game is a snoozer, so I’ll show you Lehigh’s comps, then get to the prediction.



#16 Lehigh actually has a much stronger offense than #15 UC Santa Barbara, and happens to be similar to a couple of this year’s Big 12 teams.  They should get dominated on the boards, but might be able to hold their own in terms of shooting, turnovers, and getting to the free throw line. 



This comparison to Gonzaga is a lot less favorable than the offensive one.  Lehigh has a terrible defense, and shouldn’t be able to do anything, really, to stop the Jayhawks.  If KU has a bad shooting night, the Mountain Hawks may be able to swoop in and grab the boards, but that’s about the only positive thing I can envision their defense accomplishing.  The prediction agrees:



The similarity prediction in this case mirrors the standard one.  There’s no reason to think this won’t be a ridiculous blowout, so even KU fans will likely prefer to watch the other games in this time slot.