How good is 'Nova's draw?
posted by Pete @ LetsGoNova.com3/17/2009 01:28:00 AM
Sponsor:Calling all 'Nova fans! Watch ALL March Mayhem games at Buckhead Saloon. $2 U-call-its. Book a table for a FREE BEER TOWER. We have four projector screens and 15 plasmas. Call 215.925.8170.
At first, I was thrilled to find Villanova in the East regional -- not only for the excellent geography, but also because I consider Pittsburgh to be the weakest of the 1-seeds by far. There is no 1-seed I'd rather see Villanova play than Pittsburgh, especially since the Wildcats have already beaten the Panthers this year.
Similarly, I am happy Villanova has a potential Sweet 16 match-up with #2 Duke. The Blue Devils scare me a lot less than some of the other 2-seeds, especially Memphis.
But then I started to take a look at Villanova's earlier path to the Sweet 16 itself -- particularly, its mini-tournament in one of the two Philadelphia pods.
Of course, it is a tremendous advantage for Villanova to play its first two games in the Wachovia Center, its home court away from home. But are Villanova's podmates conducive to an easy trip to Boston for the Wildcats?
No way.
In the first round, Villanova draws a feisty American team. The Eagles return nearly everyone from last year, when they took 2-seed Tennessee to the brink of elimination in the first 35 minutes of the first round.
This year, American boasts an RPI in the 70s and actually received a vote in this week's AP top 25! Does that look like the profile of a lowly 14-seed?
Villanova, favored by 17, should win the game, of course. KenPom gives American just a 10 percent chance at the upset. If the Wildcats are able to hold on, they will not get a pushover in the second round.
Waiting for Villanova will be either #6 UCLA or #11 VCU.
First, let's start with VCU, in the event the Rams are able to pull off the upset.
VCU is ranked 55th this year by KenPom. The Rams also have excellent NCAA tournament experience from 2007, when they beat Duke in the first round and took Pitt to overtime in the second.
Senior guard Eric Maynor is back for one more tournament run before heading to the NBA. The Rams also have another future pro in big man Larry Sanders.
VCU is going to be a very tough out in this tournament.
But its opponent, #6 UCLA, looks even better.
Is UCLA the best-ever #6 seed in tournament history? Well, I can't think of one better. The Bruins have been to three consecutive Final Fours, and feature three seniors -- Darren Collison, Josh Shipp, and Alfred Aboya -- who have as much NCAA tournament experience as anyone in college basketball.
UCLA also has Jrue Holiday, one of the very best freshman guards in the country -- the Kemba Walker of the West.
The Bruins are ranked #17 in the nation by the coaches, and #18 by the writers. Those are significantly better rankings than the 21-24 range implied by the #6 seed.
Most frightening, however, is UCLA's KenPom rating: #9 in the country. For a 6-seed! Ahead of 3-seeds Mizzou (#10), Kansas (#11), and Villanova (#19). The Bruins are also ahead of 2-seeds MSU (#13) and Oklahoma (#17).
That means, according to KenPom, Villanova would be better off playing any of the other 3-seeds, two of the 2-seeds, all of the 4-seeds except Gonzaga, and all of the 5-seeds, than to play 6-seeded UCLA in the second round!
In fact, since UCLA is ranked a full 10 spots higher than Villanova, the Bruins would be significantly favored on a neutral court over the Wildcats. It's a good thing that the Wachovia Center will not be a neutral site for that game.
An obvious repudiation of this argument is the idea that the tournament is tough for everyone.
Well, not this tough. Look at the other 3-seeds this year and their paths to the sweet 16.
- Missouri: Cornell; Utah St./Marquette
- Kansas: ND State; Dayton/WVU
- Syracuse: SF Austin; Temple/ASU
There is not a team on that list I wouldn't rather play than UCLA.
All in all, if the Wildcats are lucky enough to get past American on Thursday, UCLA is a monster assignment. It's more like an elite-8 game than a second-round match-up.
VCU would not be a cake walk, either.
If not for the friendly confines of the Wachovia Center, I'd be even more worried. But with the home-court edge, I think Villanova gets to Boston -- barely.
Once the 'Cats are in Boston, it's the smoothest possible sailing to Detroit. I really like the team's chances vs. Duke or Pittsburgh (or Texas or FSU).
So what do you think? Vote in the poll below.
Did the selection committee give Villanova a bit more of a test, factoring in the pseudo-home-court advantage? Or did Villanova just get unlucky in its first-weekend draw?
Maybe you totally disagree with me, and love Villanova's draw to the Sweet 16.
Either way, let me know in the comments.
Labels: american, bracketology, NCAA tournament, NCAAs, seeding, ucla, vcu, Villanova vs. American, Villanova-American
23 Comments:
Like I said, love the 2nd weekend draw, am terrified of the first.
What do you think?
American game will be close at half...we might even be losing. After that it will be all Wildcats. I saw American in a great struggle with the Cadets last weekend. They are good but in their conference. Wildcats will put them away in the last 10 minutes...going away by 15 or 20.
As for UCLA, they won't even be playing on Saturday. VCU will beat them. The Bruins have no inside game and they have played awful as of late. Either UCLA or VCU will be a challenge...I'd rather play UCLA. In a VCU game, if it got close, the entire Wachovia Center would seem to be pulling for them. ITs the strangest thing but that's what happens in the tournament. Go Wildcats!
Pete have had some issues with your stuff in the past but this is a good preview. Even if you treat the Wac as a true home court Pom's numbers would have us barley a 1 point favorite versus UCLA. For purposes of Vegas, given the American spread, they are certainly treating us like we are playing home.. though I think thats because the public might overvalue that advantage some.
UCLA is a nightmare match up for us. All this talk about how they can't score should be thrown out the window. They have a great efficiency differential in conference games this year and they are second in the nation in FG% at 49.4. Seems to be a team that certainly makes the most of their possessions on offense while playing solid D.
I will be cheering for VCU for certain. Have to think that we can handle either at Wachovia but it does not seem like it will be easy.
Just make it to Boston.
KenPom is a smart guy with some valuable insights, but I really think that his rating system has problems. In this case I'm inclined to think that the polls have it right, which means that UCLA is better than a typical 6 seed, but not that much better. I also think he has Nova way too low.
But I agree that UCLA won't be a cakewalk by any means.
first time i've ever agreed with you. i was thinking the same exact thing.
Pete-
I like how the Title of your post is "How Good is Nova's Draw", but the URL is 'how-bad-is-novas-draw' change in heart mid-stream? Nice job on the post Pete.
More on KenPom (kind of tangential to Pete's analysis, which is not bad). Does he anywhere explain his rating formula? Because his ranking of UCLA versus his ranking of Nova is just inexplicable.
Conference - identical records in conference, but Nova in a significantly tougher conference. More specifically, giving KenPom the benefit of the doubt by using his ranking numbers for other teams, in conference Nova's best wins were versus 6, 15, 15, 20, and 20 and losses were versus 3, 4, 4, 8, 20 and 26 (I'm including the tournament). OTOH, UCLA's best conference wins were versus 16, 27, 27, 29, 29 and 31 and losses were versus 12, 12, 16, 27, 31, and 39.
Out of conference, Nova's record was marginally better against similar competition (both lost to Texas (28); Nova with wins over 47 and 75, UCLA with a win over 37 and a loss to 49.
Another way to look at is is that Nova's worst loss was to the number 28th team in the nation; UCLA lost to the same team and had 3 losses to worse teams. OTOH, UCLA's best win was over the 16th best team in the nation; Nova had 3 wins against better teams.
So how is UCLA rated over Nova by KenPom? Not a rhetorical question, I'm honestly curious.
#7,
KenPom uses offensive and defensive efficiency to rate teams. It's not a traditional rating system like the AP or coaches.
That's why it's even scarier -- KenPom is way better than regular votes. He uses a mathematical model.
Ha, #6,
I figured since the two phrases mean exactly the same thing, might as well appease the "negative hater-haters" this one time and put "good."
"That's why it's even scarier -- KenPom is way better than regular votes. He uses a mathematical model."
With all due respect, Pete, the fact that KenPom uses a mathematical model does not, by itself, make his results inherently credible. Nor is KenPom the only person out there with mathematical model. Other models support a more "conventional" ranking of the two teams.
A mathematical models can be flawed in several ways. Most fundementally, a model is only as good as its underlying assumptions. Garbage in, garbage out. From what you say, the fundemental assumption underlying KenPom's model seems to be as follows: "offensive and defensive efficiency are better predictors of a team's performance than wins and losses, even when we take the quality of the wins and losses into account."
I would say that this underlying assumption is subject to dispute. At the very least, I'd like to see some evidence that it is true, especially since KenPom tends to depart pretty significantly from the conventional wisdom.
And I'm not saying that KenPom has a bad track record in terms of predictions. But anyone with a decent knowledge of the game can predicts wins and losses mre than 50% of the time. How does KenPom compare to more conventional methods? What is the data?
I am not here to get into a debate on KenPom's record of success or lack thereof -- I don't have that data.
But I do understand that a model is only as good as its input -- and I trust KenPom knows what he is doing.
Efficiency-based statistics are revolutionizing every team sport as we speak -- baseball, basketball, football. I put a lot more credence in efficiency numbers than I do in traditional polls that have more to do with "who lost last" than anything else.
But, of course, we will see what happens. The one good think Nova has going for it over UCLA is the Wachovia Center. I have Villanova reaching Boston.
Well Pete, a farily reasonable response; my only real quibble is that we aren't merely talking about "traditional polls," but also more sophisticated models which are based upon wins and losses (taking into account opponents). The real issue isn't mathematical models versus subjective analysis; it's whether winning and losing is better predictor of winning and losing than efficiency statistics. I suspect the former is more accurate, but I'll admit I don't have the data to back it up either.
Which isn't to deny that efficiency statistics have their role. For example, efficiency statistics tell us that, somewhat contrary to popular impression, UCLA has a darn good offense. But do those stats "prove," or even indicate, that a team with a (slightly) worse record against a (somewhat) easier schedule is better than a team with a better record/tougher schedule? Color me sceptical.
Since it's basically a home game, why aren't the Nova fans turning out? I have tickets on Ebay - good ones in section 103 - that have not sold. I'm shocked!
Probably because you are charging too much.
12,
good points.
Lots of Ken Pom talk. I'd say his numbers are generally a good measure but #19 vs. #9 is not all that important. UCLA is a good team and I think everyone recognizes it is gonna be uncomfortably close and it is definitely not the easiest of 6 seeds to face. However, you cannot compare 2 of the other 3 seeds brackets (Big12 guys) because they are matched up against Big East teams (Marquette/WVU) and the committee would never set it up so Nova would face those guys that early. Maybe Cuse's Pac10 matchup is slightly easier in ASU, but that's probably a toss up. As for American...well if Nova loses to them in Philly than the bracket talk means next to nothing, at least to me, because they were going nowhere no matter who they were matched up against.
The bracket is a good draw for Nova because if they are to make a deep run in the tournament then they have the best set up they possibly could have asked for. As much as I would love an easy road to the Sweet 16, that almost never happens so I'll gladly accept the more favorable road to the promised land that is the Final Four.
But that is just my opinion...
Kenpom has proved to be extremely accurate in the past. Do a quick google on kenpom and last years tournament and you will see his system came in first place by a large margin against a pool of "experts" and other mathematical models. As a side note most other popular mathematical models have UCLA ranked higher than villanova also.
Also, all his number are adjusted for opponents schedule strength. His blog has a good explanation of this.
About UCLA: I think that their best is much better than nova's best. However, they have a lot of freshman and are fairly inconsistent. That being said Howland will probably have them ready to play. I'm sorry to say but I have a bad feeling about this one...
first post that didn't bash on jay wright in a long time......making some progress cilento!
#17 - We all agree that UCLA will be tough. But I've looked a bit more closely at KenPom's methods as applied to UCLA and Nova, and the basic reason that he has Nova and UCLA basically reversed compared to almost everyone else* is that UCLA has a bunch of blowout wins versus bottom rung teams, while Nova tended to some extent to play down to the level of their opposition when they played lousy teams (but still won). If one looks at the efficiency numbers against decent teams - NCAA Tournement caliber teams - Nova has the edge. That's true however good KenPom does generally at predictions (along those lines, it would have been nice to see a link to support your assertions; in any event, one NCAA tournement is a pretty small sample size).
*and you are just flat out wrong when you say that UCLA is ranked higher than Nova in "most other popular mathematical models."
http://i.gizmodo.com/5173667/ibm-wants-sun-for-7-billion
http://forum.dvdtalk.com/other-talk/405839-damn-petite-teen-swallows-11-pounds-big-beef-record-time-wow.html
Pete,
Your picture of Badlands is an internet meme. Congrats.
#19 Let me clarify. The most common mathematical models I see cited are Kenpom, Sagarin, and LRMC. Kenpom and LRMC have UCLA well above Nova, Sagarin has Nova 2 spots ahead of UCLA in his Elo ratings, but in his predictor ratings (which he claims are more accurate) UCLA is well above Nova. I think "flat out wrong" is a little strong.
-Both UCLA and Nova have many blow out wins over crappy teams. I count 14 or so for Villanova and 14 or so for UCLA. Kenpom has UCLA and Nova at approximately the same SOS. In fact if you remove all of UCLA's blow outs over crap teams their adjusted offensive efficiency actually improves (this is because kenpom adjusts efficiencies based on opp. strength). Removing Nova's games vs god awful teams keeps them at about the same place. So I don't think your blow out/plays down to opponent theory has much weight.
-It is true that Nova has slightly better efficiencies vs teams that actually made the tournament than UCLA. But I think its kind of subjective to use that measure (I find the above example of removing crappy teams subjective as well).
-I admit I haven't been able to find pre-2008 tournament kenpom results, but I find his predictions to come in at about 70-80% accuracy. However, talking about prediction accuracy is a little disingenous as he is really only giving a probability that a certain team will win. Here is the link that you asked for: http://blogs.wsj.com/numbersguy/ken-pomeroys-winning-bracket-316/ (click on the google spreadsheet in the article)
Go Nova!
Great article in today's Daily News on Reggie Redding....its a must read.
@22
here is the Redding article --
http://www.philly.com/philly/sports/colleges/20090318_Redding_accepts_his_role_for_Villanova.html
Post a Comment
<< Back to LetsGoNova.com