For the Ken Pom Guys

How come the SEC is rated so low…(or is that just my biased view)?

His ratings are based on a formula. They are different in that they are not solely results based. It measures how you play, too, not just the end outcome, which is what the RPI measures.

The eye test, road record, last 10 games and every other reason to justify a seed. It all stinks. I keep hearing how great the Big 12 is in my opinion the SEC is a tougher conference and better! If there a way for the pundits and committee to leave the hogs out they will!

Why do you think the SEC is rated low?

Number of teams in Ken Pom top-40:

ACC - 8
SEC - 7
Big 12 - 6
Big 10 - 5
Big East - 5
AAC - 3
P12 - 2

I think the SEC is getting plenty of credit. I expect seven or eight teams to make the tournament and that’s more than I’d expected before the season.

Winning will take care of it all! I’m already Tired of lunardi, The polls and Pom.

Me too. He have a heckuva a country where a guy can make a good living making guesses about a game.

The SEC’s 4 loss or 8 loss D1 teams don’t seem to line up with those of the rest of the conferences. For example Cincy is #5 and Auburn is #10. This greatly affects tourney seedings and therefore likely tourney wins and losses. WVA is #13 while our Hogs are 239 (or ish).

What does it mean by “measuring how you play”. “Measure” and “How you play” together sounds oxymoronic.

I think you’re cherry picking here. A&M is 17 - 10 (#26), and nobody with a worse record is rated higher. Likewise, there are plenty of teams with better records than UT that are rated below them. There are also plenty of power conference schools with the same or better records than Arkansas rated below us like KSU, which is also in the B12 like WVU. The individual Ken Pom scores will depend on SOS, which also includes nonconference schedule, and margins of victory will weigh heavily as well. Hence, record and ranking between any two teams isn’t necessarily reflective of conference power. Blowing out opponents and losing by small margins will give a higher ranking over a team that has the exact same wins and losses but did the opposite.

Also, the numerical ranking distorts the difference between teams. For instance, there is as big a difference in the raw score from #1 to #10 as from #11 to #63. Quite frankly I think that represents a bug in his algorithm. I don’t believe it and don’t consider the ranking to have fine resolution, which is why looking at the number of teams in a certain range is as meaningful a measure of power as any. You can tweak the formulas and get a lot of shuffling, and there aren’t enough games to conclusively say one formula is better than another.

I prefer a poll of polls like the Massey Composite where Arkansas is #29. I don’t believe any single computer poll has a monopoly on ground truth. The computer polls in general tend to overrate the Big 12 every year because it is a small conference with no lousy teams. That creates feedback effects in the SOS that often exaggerate how good their best teams are.

Pomeroy quantifies the kind of things that Jimmy Carter used to write about here, and Scottie does now. Offensive efficiency, defensive efficiency (both measured in points per 100 possessions), tempo, efficiency margin over your opponent, strength of schedule (but much different than RPI SOS), and even luck. There was one recent season (when the Hogs were good) that we were the absolute unluckiest team in the country. Pomeroy believes that how matter how good a team is, its record in close games should be about .500, and any deviation from that is luck. He also quantifies other things that I can’t see because I don’t pay to see his more detailed numbers.

But if you score 114 points per 100 possessions and only give up 82, you’re gonna win a whole lot of games. Those are the current Pomeroy numbers for Virginia, which is the only team in the country with only two losses. Our offensive efficiency is a little better than UVa, but our defense is just about 1.00 ppp, so that’s why our record isn’t as good as the Wahoos. But 1.00 ppp is 83rd in the country, not terrible but not the level of defense you need to win an NC in recent seasons. It can happen, though; Duke entered the 2015 tournament with a defensive efficiency ranking in the 50s and came out of the tournament with a ranking of 12th – and an NC trophy.

I just got curious and tried to do my own Pomeroy numbers for the '94 NC team. He adjusts numbers according to your schedule, which I have no way to do, but I can do raw numbers, and what I found surprised me. The defense played by our heroes was really not very good. We gave up 1.091 PPP, even considering the number of times we turned people over. This year 1.091 adjusted PPP would be 241st in the country. But our offense? Almost off the charts, 1.285 PPP. The only team in the country this year better than that is Villanova.


I think you’re cherry picking here.

[/quote]Thank you. I guess it appears like I cherry-picked, but I picked my school and the team in our division that’s at the top. In my opinion, the SEC is hands down better than Cincy’s conference, yet our conference’s leader (with the same record) is 5 positions south. On the other hand, WVA with the same record as the Hogs, yet trending south while the Hogs are trending north, is way higher than the Hogs. Just a bit incongruent to me.

Thanks. Makes sense. Good info.


I agree. I don’t believe that Cincy is all that, although they are obviously a quality team. UF beat them on a true neutral court. If you look at their results, one thing that the Bearcats did was keep the pedal to the metal against some of their weaker foes like a 50-point win over Western Carolina. In Pomeroy that helps mitigate that WC is a lousy team on the SOS. His formula may predict that Cincy should win by, say, 30 points. So, they can still gain ground against weak teams in polls that take into account margins by beating expectations. By contrast, you just get a W in RPI and take a hit on your SOS for a game like that. RPI has Auburn ahead of Cincy for that reason. I much prefer Ken Pom to RPI, but Ken Pom has distortions of its own.

It’s not good sportsmanship but playing to the buzzer can make a difference in the computer polls when you have a sizable lead. For that reason it aggravates me that we often give up a few late treys in games that have been decided. Likewise, not giving up can make a difference if you are on the other side of that equation. Losing by 8 is better than losing by 15.

The reason Pomeroy does things the way he does is that many stats are tempo dependent. A team that plays like we do is going to score more points (and give up more) than a team that plays like Virginia. The Cavaliers have the absolute slowest tempo in the country, 59 possessions per game. We average 71. OU averages 76. Savannah State averages 81 but only scores 1.03 PPP. If you do the math, the average possession in a Savannah State game is less than 15 seconds. The average possession in a Virginia game is a bit over 20 seconds. Doesn’t sound like much of a difference, but it is a huge one. Remember how slow Stan Heath’s style of play seemed? His last Arkansas team averaged 66 possessions. As noted above, we average 71.