A little chit-chat about KenPom

[Rhavicc]

First and foremost, UK has restored order. We're #1 in the country according to KenPom again (Not sure how we fell off honestly), we rank #1 in the country in defensive efficiency by far, and we now rank 6th in the country in offensive efficiency.
It looks like Camp Cal did us a world of good offensively after all, seen a lot of excellent ball movement yesterday. Helps us a lot that we're hitting our shots as well. Now if we can keep consistent aggressiveness in the paint, and the Harrison twins let the game come to them like they did against Alabama, all will be well.
That said, somebody riddle me this. KenPom has our out of conference SOS ranked 83rd, and our overall SOS ranked 30th. Would anybody care to explain this to me, because I'm lost when it comes to that stat, unless the SEC became America's top conference by a mile, overnight.

 

[bigbluelou]

Originally posted by Rhavicc:
First and foremost, UK has restored order. We're #1 in the country according to KenPom again (Not sure how we fell off honestly), we rank #1 in the country in defensive efficiency by far, and we now rank 6th in the country in offensive efficiency.
It looks like Camp Cal did us a world of good offensively after all, seen a lot of excellent ball movement yesterday. Helps us a lot that we're hitting our shots as well. Now if we can keep consistent aggressiveness in the paint, and the Harrison twins let the game come to them like they did against Alabama, all will be well.
That said, somebody riddle me this. KenPom has our out of conference SOS ranked 83rd, and our overall SOS ranked 30th. Would anybody care to explain this to me, because I'm lost when it comes to that stat, unless the SEC became America's top conference by a mile, overnight.

Two thoughts:
The OOC SOS is tainted by some of the terrible teams. Like we've argued, it's silly to give equal weight to the 175th team, the 225th team, the 275th team and the 350th team. They all should be given much less weight than the ones in the Top 100, Top 50 and Top 25.The SEC is actually pretty highly rated top to bottom. We may not be as top-heavy as, say, the ACC, but we have a lot of teams in the 25-90 range.

 

[UKnCincy]

First, almost every team from a power 5 conference will have a higher ranking in overall strength of schedule compared to their non-conference strength of schedule.

As far as UK is concerned, there are lots of small schools that play their non-conference schedule against power 5 teams. If one on those teams is playing 5 or 6 of their 13 or so games against power 5 opponents, that's going to be pretty brutal and would rank them ahead of UK when comparing non-conference strength.

However, a team with that brutal of a schedule still will play the majority of their schedule against their own conference. If more than half the time they're playing Horizon league teams and more than half the time we are playing SEC teams, then you're going to see them swap places when you compare overall schedule strength.

 

[fatguy87]

Our average conference game is more difficult than our average nonconference game. In the noncon, we played the likes of Boston, Grand Canyon, and Montana St. These teams are well below average on a neutral court, and even worse than that when playing in Rupp. There aren't any teams in the SEC nearly as poor in quality as those teams.

We also play road games which greatly increases the difficulty of a game relative to a home game. The game @Alabama, for example, was considered more difficult by all objective systems and Vegas than every game on our schedule but the game @UL (Kansas excluded since they appeared to be heavily overrated by preseason prognostications). Even wins against mediocre teams can be difficult. The Vegas line for @A&M looked like the line for our games against Texas and UNC. There is a disconnect between the actual value of homecourt advantage and how the general public, media included, evaluates wins.

 

[SilentsAreGolden]

I've learned to stop even trying to figure out SOS rankings. They make ZERO sense. As many other have pointed out, teams from about 150-350 are effectively about the same. If you beat #308 by 60 or team #159 by 35, is there really any difference? Teams that don't play anybody above 150, but also don't play anybody in top 10 will be ranked ahead of teams like us who play several teams in the top 10, but a few cupcakes that are in the 200 or 300 range. It's moronic. This fact makes the SOS argument moot as hell. No value to it whatsoever.

 

[JonathanW]

I am a statistician, so analyzing data such as what KenPom does is what I do for a living.

I don't know about his SOS calculations, but many simply take an average of the ranks of your opponents. This method is inherently flawed. I can best illustrate this with an example. Say Team A plays vs teams ranked 130, 140, 150, 160, 170; and Team B plays vs teams ranked 10, 20, 310, 320, 330. Team A opponents average rank 150, whereas Team B opponents average rank 198. Who played the tougher schedule? I think it really depends on where Teams A & B rank. If they rank say in the top 50, then Team B played the tougher schedule, having played 2 teams better than them and 3 automatic wins, whereas Team A played all 5 games vs teams they should probably beat. But if Teams A & B rank 110 or worse, then Team A has the tougher schedule with all 5 opponents with a reasonable chance of beating them.
SOS should be calculated relative to your own team's ranking, but in most statistical models is not.

 

[Rhavicc]

So, summing it all up, the SOS system is broken and I should just leave it to the talking heads at the selection committee (though I doubt we'll need one for us this season)

 

[wild mandu]

Like any computer system, Kenpom is a little wacky. Do you honestly believe San Diego State has a better defense than Virginia?

Whether it's good or bad for your team, I've learned to mostly ignore stats and computer rankings. My eyes tell the tale. Alabama would have another football title if we left it up to that nonsense.

 

[Vandalayindustries]

I don't know a lot of details but I know that Grand Canyon, Montana St, and Miss State are REALLY bad.

 

[BoulderCat]

I believe the SoS associated with some of our lessor opponents is what brought down UK's overall despite obviously having played a number of games against top 10 and 25 teams. Teams like Grand Canyon. They are 2-0 in their conference and 11-8 overall, but they probably started at a very low SoS, and have played other teams like them. Cal and UK are nice. bring them in for a game, and it hurts us in the stats. I'd rather be nice myself. I think it was Lumbardi I heard yesterday say KU is a good example of a team that always has high numbers. They don't play any/many teams below the 150 or so ranking is why. I could care less if their numbers are better, we humiliated them, and we're undefeated and ranked #1.
This post was edited on 1/18 3:28 PM by BoulderCat

 

[BigBlueFanGA]

Originally posted by JonathanW:
I am a statistician, so analyzing data such as what KenPom does is what I do for a living.

I don't know about his SOS calculations, but many simply take an average of the ranks of your opponents. This method is inherently flawed. I can best illustrate this with an example. Say Team A plays vs teams ranked 130, 140, 150, 160, 170; and Team B plays vs teams ranked 10, 20, 310, 320, 330. Team A opponents average rank 150, whereas Team B opponents average rank 198. Who played the tougher schedule? I think it really depends on where Teams A & B rank. If they rank say in the top 50, then Team B played the tougher schedule, having played 2 teams better than them and 3 automatic wins, whereas Team A played all 5 games vs teams they should probably beat. But if Teams A & B rank 110 or worse, then Team A has the tougher schedule with all 5 opponents with a reasonable chance of beating them.
SOS should be calculated relative to your own team's ranking, but in most statistical models is not.

So the #1 team is always at a disadvantage in your SOS model?

 

[wildcatsboston1984]

Originally posted by Rhavicc:
First and foremost, UK has restored order. We're #1 in the country according to KenPom again (Not sure how we fell off honestly)

Because our offensive efficiency fell to 12th at one point while UVA was top five in both offensive and defensive efficiency, and with a better Pomeroy SOS.

 

[SaguaroCat]

His sos values sound too volatile.there's no reason that playing four teams as mediocre as ole miss, Texas am, Alabama and Missouri would increase our sos from 80 to 30.

Sagarin would appear to make more sense as it gives us the 18th sos. Not sure what it was before conference, but it was closer to 20 than it was to 80.

 

[SilentsAreGolden]

KU has it figured out I guess. They are always in the Top 5 or so in SOS and RPI. Remember last year when they were still in the Top Ten with 10 losses? SOS does you a world of good in these models.