I’ve been promising all season that I’d crank out a quality curve comparing the efficiency of this year’s seeds to their historical counterparts. Up until now, I’ve just taken the KenPom top 20 with no sense of where they would be seeded. Of course, the NCAA Selection Committee never follows Ken to a tee. So I wait until the bracket is out before comparing the efficiency of this year’s seeds to like seeds from the past nine dances.
Well, I just did the curve—and the results are, in a word, scary. Take a look at what we have:
Top seeds are close to the average quality of one seeds. But with each succeeding seed from two to six, the teams get progressively weaker compared to their historical strength. Sevens and eights are a slightly better than average, nines and tens are solidly weaker…and 11’s and 12’s are significantly stronger. As for the last four seeds, 14’s and particular 15’s are much stronger.
What does it all mean? Here are my first-blush thoughts:
- Let’s see—really weak five and six seeds and stronger 11 and 12 seeds. I smell more upsets. UNLV, UCLA, Butler are particularly weak high seeds. St. Mary’s, Minnesota and Ole Miss are unusually strong low seeds.
- Super-strong 15 seeds and meh two seeds? Uh-oh. Don’t tell me we’ll have another teeth-gnashing bracket crusher right out of the chute? Georgetown and Miami are the least efficient two seeds. Pacific and Iona are the most efficient 15’s. Iona is head and shoulders above the other 15 seeds. Alas, they play the strongest two seed in Ohio State. Pacific, on the other hand, gets Miami. Hmm. I say, “Hmm.”
- If sevens are strong and twos are weak, could we be looking at some second-round 2v7 shockers? If so, the powerful sevens are Creighton and San Diego State. Creighton gets a solid Duke squad, but Fisher’s bunch gets a vulnerable Hoya team.
- With the ones close to average quality and everyone else pretty weak, we could be looking at waltzes to the Elite Eight. But watch out for one eight seed in particular. Pitt is the seventh most efficient team in the land. They would take on a Gonzaga team that’s slightly weaker than your garden-variety one seed.
Anyone seed anything else that jumps out?
Peter Tiernan has been using stats to analyze March Madness for 22 years. He writes for 
Quick question on the upset/toss-up model (for making the actual model) . . . when an upset qualifies, does the higher seed need to also qualify as likely to underachieve in order to move the lower seed on . . . also, I noticed last you you moved on two 13 seeds, though the percentage for upset is lower than .500 (though the model could have changed slightly since last year)? Basically, I am asking if you account for higher seeds at all (in upset matchups) when filling out this model
In upset/toss-up, I don’t think I accounted for higher seeds in 5v12 and 6v11. Seem to recall that I did for 4v13. You could try it both ways and just document what you do. Use the same format, Ryan, as I gave you before. Haven’t had a chance to look it over. But will try to get to later this afternoon.
Thanks again, my man.
ok, I will send the final product your way later tonight, I want to recheck everything . . . as of now, how I did it, it still produced some interesting results (like Notre Dame in the Final Four????)
I made a correction on that last comment, haha. You will see in the email.
Nice work on the curve . . . I checked this out last too and was a bit surprised how all over the place it is. 11 seeds are actually better than 6 seeds (on average). 8 seeds are much stronger than the 7′s and 6′s (and comparably strong to the 5′s). The only five seeds stronger than average are the 8, 11, 12, 14, and 15 seeds. Should make things interesting.
Seeing that Quality Curve (or to better phrase it…the roller coaster of inadequacy) makes me want to step back 10 feet and start throwing darts at the bracket. I think my dog has a better chance of picking games.
Hahaha.
Reading the exchanges between you guys (Ryan and Pete) makes me want to be in the bracket science inner circle
If we got together to talk statistics I’m pretty sure no one would ever leave!
On crazy days like this, Andy, I often wish I was on the outer circle!
I know someone was working on the upset/toss up model is that up anywhere?
Coming soon…with the second wave of models.
If you ask me, the easiest “upset” pick of 11 over 6 is Minnesota over UCLA. UCLA lost a great player do to injury and their star Muhammad is too immature and selfish (check his assists stat). Right now, he is going to put UCLA on his back just to improve his draft stock (like Barton did in last year’s Memphis squad). I am more concerned about the 3 vs 14 seeds. New Mexico has the easiest task (a watered-down Harvard), but Florida and Marquette does not.
Agreed on Marquette vs. Davidson. Heck, anyone could come out of that Butler|Bucknell|Marquette|Davidson quadrant. On Minnesota, they have historically bad momentum. Only one other team made the dance with 3 wins in L10 and a three-game pre-dance losing streak. That was Villanova in 2011…and they lost in the first round.
Hey Pete,
Love the work, doesnt’ this year’s top 20 Pythag match up almost identical to 2011? Also, it seems that is closer to 2011 then 2012. I bring this up, because in 2011, we only had 2 upsets for teams seeded 1-5, but three 6 seeds lost in the first round. But in 2012, we had 5 teams seeded 1-5 lose and two 6 seeds. So based on some of your info, doesn’t it seem that the first round of this years dance could follow suit, ie (pretty tame first round and then craziness starting in round two)?
Just a correction, I believe that Ohio St was rated the weakest of the 2 seeds.
I’m not looking at what the committee said. I’m looking at efficiency numbers. Of course, KenPom says Florida is the best team in the country. So…
Why Minnesota? 3-7 in their last 10 pre-tourney games. That doesn’t scare anyone else? Is UCLA’s loss of Jordan Adams that devastating? Is there something else I’m missing?
Efficiency numbers say Minnesota is better than recent performance. But as I just responded to the earlier post: On Minnesota, they have historically bad momentum. Only one other team made the dance with 3 wins in L10 and a three-game pre-dance losing streak. That was Villanova in 2011…and they lost in the first round.
I love the vs-avg perspective. What I’m really curious about is the particular year in the past to which this year’s curve conforms.
I also wonder what KenPom’s numbers would look like from a relativity calculation. Florida may be #1 in possession based stats, but half of FLA’s possessions came against the SEC whereas half of INDs possessions came against B10. Seems like possessions vs SEC could inflate FLAs numbers better than they should be.
KenPom data does factor in strength of schedule.
Based on kenpoms #’s, two near toss up game/top seed knock offs in round 2 in the West could be Pitt over the Zags and Arizona over New Mexico.
Lou – I tell you, that Pitt squad is dangerous. And New Mexico gives me the yips. Steve Alford is one of those infamous snakebit coaches I always talk about. So is Few…and Brey. In fact, the only big coaching overachiever in the whole bracket is Sean Miller. Maybe the committee took pity and lumped all the bumblers together so one might win.
Hey Pete, just wondering if those 11 12 seeds you mentioned (St Mary’s, Minnesota, and Ole Miss) all possessed the traits in your “seed guide to upsets?” I’m assuming they do if you considered them “unusually strong.” Just wanted to be clear. Also will you post anything identifying which teams possess all the victim or upset qualities? No biggie if you don’t, I like doing the research too. By the way, awesome job once again.
I based those comments on KenPom value proximity of the upset seed to the victim seed.
How about the potential 10 seed Colorado over 2 seed UMiami to advance to the sweet sixteen?
Miami (Fl) is tagged as a potential underachiever in the upset rules
Jim Larranaga is a fantastic coach though (he loves stats, btw, one of the leaders in the field for college coaches)
Agreed Tressler, I feel that UMiami achieved their goal (ACC champion), but they do not have a Final Four talent. Sweet sixteen, maybe. I believe Colorado has the firepower to go to the sweet sixteen, or am I missing something?
I just penciled Colorado in for a Sweet 16 run in the Contrarian bracket.
Okay…I have gone through the seed guide and upset spotting rules and I am struggling to find any “confident” victims in the 5v12, 6v11 games like I saw last year. Anyone else seeing this?
Nick – upset rules aside, take a look at the Seed Quality Curve I did earlier today. Take a team like Minnesota. Momentum numbers say steer clear (3w in last 10, 3 game losing streak)…but the Gophers have pretty solid efficiency numbers. Will they thrive not having to play another Big Ten team?
I was very disappointed that neither came out with a 5 v 12 upset this year. However, I will say this, Oklahoma State and Wisconsin are vulnerable to upsets according to the Upset Tips. Then if you take into consideration that Oregon and Ole Miss are both under-seeded according to Pomeroy’s ratings (Oregon should be a 10 seed, Ole Miss should be a 9), I could easily see an upset coming out of those two pairings. As for the 6 vs 11 match-ups, the upset/toss-up model does have Belmont and St. Mary’s not only pulling off first round upsets, but also second round upsets. In fact, it has Belmont going all the way to the Final Four
Excellent feedback. Cheers!