Wednesday, March 21, 2012

Sweet 16 Open Thread + Picking The Lines

Well, I hope nobody bet their life savings on my picks against the spread this year. It's been a brutal run of late, and it's going to take a big turnaround to finish the Tournament over .500.

That said, I do still have all 8 of my Elite Eight teams left in my bracket, with all but one of the teams I picked (Louisville) favored in Vegas. And when people at ESPN are making predictions like this, it makes me feel better that at least my down years aren't anything like that.

I am glad to have the NCAA Tournament back in my life. I've been in a bit of withdrawal lately. When the Sweet 16 tips off, please join me in the comments to this post and on twitter.

Sunday ATS: 2-6-0
Total through Friday ATS: 23-27-2
2011 Tournament ATS: 40-26-1
2010 Tournament ATS: 35-25-3

Wisconsin (+4) over Syracuse: This is actually an awfully difficult game to pick. Both teams match up very well against the strength of the other team. The Syracuse offense is very dependent on transition baskets off of turnovers. They tend to struggle in their half court offense and are a bad rebounding team on both ends of the floor, but they might be the most dangerous team in the nation at transition offense. But Wisconsin? There's probably no team in the nation better at forcing a deliberate tempo, preventing turnovers and getting back on defense than the Badgers. At the same time, Wisconsin's offense tends to struggle against the zone. By having their big men patrolling the perimeter they were able to take Vanderbilt's Festus Ezeli out of the paint, opening things up for their guards. That strategy won't work against the Syracuse zone, which is going to turn Wisconsin into a perimeter shooting team. And the Badgers are a team that tends to live and die by the three. This season they have been 17-0 when shooting 38% or better behind the arc, and 6-8 when shooting under 36%. If they can't hit their threes against Syracuse, it won't matter if Syracuse can't score, because Wisconsin is still going to lose. I do think that there's a good chance that this game comes down to the final possession, so I'll take the points.

Louisville (+5) over Michigan State: This is an upset that I picked in my original bracket breakdown here. Michigan State is a better team, but I think Louisville is a match-up nightmare for them. Michigan State is sloppy with the ball (8th in the Big Ten in offensive turnover rate), and they have a tendency to allow themselves to get sped up by teams that want to run. The Louisville press has been outstanding the past month, forcing turnovers on more than 20% of possessions in 7 of their last 9 games. Louisville's biggest flaw is defensive rebounding, but Michigan State hasn't been strong on the offensive glass since Branden Dawson (their best offensive rebounder) got hurt. Louisville is my pick to win this game outright.

Cincinnati (+7.5) over Ohio State: I certainly don't think Cincinnati is going to win this game, but this spread is awfully large. Both of these teams are very good defensive teams. Ohio State is rated the best defense in the nation by Pomeroy. While Cincinnati is only rated 22nd best, their defense has been at its best the past few weeks. They completely shut down Florida State in the Round of 32, which allowed them to overcome mediocre offense of their own. The biggest concern I have for Cincinnati is front court depth. They don't have the array of bigs that Gonzaga has. If Yancy Gates gets into early foul trouble, this game could get out of hand. But if he can stay in the game, I can see this being a real slog, so I'll take the points.

Marquette (-1.5) over Florida: Florida blew out its first two NCAA Tournament opponents, but their strength of schedule has been awfully weak. Virginia finished the season in a tail spin and had only beaten one NCAA Tournament team (NC State) since November. Meanwhile, Norfolk State was who we thought they were - the third best team in the MEAC that simply shot the lights out in one crazy, fluke game against Missouri. Florida's defense is shaky, and they haven't beaten a team as good as Marquette all season long. Florida doesn't turn the ball over a lot, but they will launch a ton of threes. If those aren't falling then Marquette will turn those long rebounds into transition opportunities. Heck, Marquette turns made baskets into transition opportunities. That will tax a Florida team that normally expects heavy minutes from its starting backcourt, and that prefers to play at a much slower pace. I just don't see the argument for Florida here.

45 comments:

Sean said...

It is really funny to see how wrong that ESPN guy was. 0/10 (unless you still give him a shot that Michigan State could be the first one seed to fall, but he made an argument that they'd lose to Memphis).

Jeff said...

Yeah. What I find irritating is when people don't own up to it. He's out with a new set of predictions, without acknowledging how bad his last set of predictions were.

I'm having a bad year against the spread, and I'm honest about it and will try to learn from my errors to do better next time. If you don't acknowledge when you make mistakes, you'll never correct them.

ervinsm said...

looking at the MSU/Ville game, I think that MSU is really a nightmare matchup for Louisville.


I dont think Behanan/Dieng matchup physically at all with Payne/Nix on the interior. At least Behanan is physical, and while dieng has gotten better, I think hes gonna end up pushed around inside the basket.

I know no one is gonna have to matchup 1on1 with Green bc they'll be in zone a lot, but I dont see how they contain him either.

I think Appling and Siva are pretty much one in the same and if appling can be ~= to Siva its gg Lville. Super quick, not great shooters, elite penetrators and pesky defenders.

The only way I really see louisville winning this game is forcing a ton of a turnovers (not impossible) and having Kuric and the Smiths get lights out hot from 3.

I don't think either coach has an edge as these are two of the best ever for a ton of reasons.

Then we add on how anemic Louisville has been on the offensive end, its kinda hard to think they are gonna score against an elite defensive squad w/o forcing TO's for easy buckets.

Jeff said...

Certainly if this is a half court game like you describe, with big men posting up other big men all game, then Michigan State will win. But in general, Michigan State gets out of that type of play when they are against a team that tries to run on them.

I think this game is going to come down to the number of possessions. If it's a high possession game then Louisville will probably win, and if it's a low possession game then Michigan State will probably win.

Rainmaker203 said...

I wouldn't sweat having a bad year against the spread. You've had several good years in a row before this year, and since you understand statistics, you know that one bad year following several strong years is more likely attributed to bad luck than anything else, especially with the fairly small sample size that is the NCAA tournament.

If you want a better indication of how well you do against the spread overall, maybe you should start picking regular season spreads in games involving the power conference teams. No explanation for the picks necessary, just make a post the night before the games with your spread picks. Goal should be being correct enough to make a profit. Since Vegas pays -110 ($100 for every $110 bet) against the spread I guess that would be ~55%? You can get spreads the night before the games at vegas.com.

If it's too much work I understand. I just think it would be something interesting to try out in the future, especially since most of us who read your site like the math/statistics side of the sport.

Jeff said...

Most Saturdays this past year I ended up talking about a bunch of the Vegas lines on my twitter before games tipped off, but I wasn't keeping rigorous track of how I was doing.

The only reason I'm a bit cautious about too much stuff with the Vegas lines is that I don't want this to turn into a gambling website. But doing a quick preview on each Friday night and picking 5 or 10 games against the spread might be something I'd do next season. I'll see how I'm feeling about it when the next season tips off.

Anonymous said...

This is more looking back at last week than looking forward, but I had to ask. If you search "biggest NCAA upset" on google, you'll find plenty of posts that claim Lehigh over Duke is the biggest upset in the history of the NCAA Tournament.

Do you agree with with me that's just a ridiculous assertion? I even read an article comparing Lehigh over Duke to the Miracle on Ice, which made me cry with laughter at the stupidity of it.

Frankly, I think Norfolk State over Missouri has a much better case of being the biggest upset in the NCAA tournament than Lehigh over Duke. Thoughts?

Jeff said...

You're right. Arguing that Lehigh over Duke is even one of the 20 biggest upsets in NCAA Tournament history is laughable. Lehigh was only a 12 point underdog - Norfolk State was a 21 point underdog. So Lehigh wasn't even the biggest upset that same day.

But it's the same logic that I talked about a lot a few years ago when Appalachian State beat Michigan in football a few weeks apart from Stanford beating USC. Appalachian State was the best team in I-AA, which made them better than the majority of I-A teams. Stanford over USC was objectively the much better upset. But "Appalachian State" sounds like an inferior team to Stanford, so that was the one people still talk about.

So Duke sounds like a much better team than Missouri, but the reality is that Missouri was the better team this season, and Lehigh was much better than Norfolk State over the course of the season.

Anonymous said...

Good call on Louisville beating Michigan State. Shoulda listened to you.

Anonymous said...

Here's the thing,

You say:

"Syracuse hit a higher % of 3s than Wisconsin, despite WI being the better 3P% team this yr. So 3s weren't why the game was close"

Yet you don't disclose the fact that it's MUCH less likely for a team that shoots a seasonal 36% (Wisconsin) to hit 52% from three than it is for a team that hits a seasonal 34% (Syracuse) to hit 55% from three. People can make stats say whatever they want I guess...

Syaracuse hit the higher %, but Wisconsin by far accomplished the greater outlier. STATS DON'T LIE.

Jeff said...

"Yet you don't disclose the fact that it's MUCH less likely for a team that shoots a seasonal 36% (Wisconsin) to hit 52% from three than it is for a team that hits a seasonal 34% (Syracuse) to hit 55% from three."


Wait, what?

I'm hoping that you just had a typo there.

Anonymous said...

My previous post (directly above this one) is predicated on the fact that it is a MUCH GREATER outlier for a 36% team to shoot 14-27 than it is for a 34% team to shoot 5-9. It's simple math.

Therefore, comparing Syracuse's and Wisconsin's 3pt shooting as if they are equal, or arguing that Syracuse's 3pt shooting was somehow more impressive, is rediculous in a statistical sense.

Jeff said...

It's not as big of a difference as you think. Also, Wisconsin has the best three-point defense in the nation, while the zone is always vulnerable to three-point shooting, so you would expect Wisconsin to shoot better-than-average and Syracuse to shoot worse-than-average behind the arc.

Anonymous said...

Again, I left out the shot attempts (an important factor) in the previous post.

Anonymous said...

The difference between Syracuse's and Wisconsin's 3pt% defense is about 2%. So in one game I wouldn't expect such a large variation one way or another.

Jeff said...

What matters with three-point defense is not percentage but number of attempts. Ken Pomeroy had a nice study this year showing that 3P% defense is mostly luck, but 3PA/FGA defense is a very consistent metric.

Wisconsin is third in the nation in 3PA/FGA defensive ratio, while Syracuse is 282nd - 14th in the Big East.

You have to expect Syracuse opponents to launch a ton of threes. Hitting more than 4 or 5 threes against Wisconsin, on the other hand, is a very good accomplishment.

Anonymous said...

Here's a straight up question:

What's more likely, a team that avearges 36% from three shooting 14-27 against a team that gives up 31.5%

OR

a team that averages 34% from three shooting 5-9 against a team that gives up 29.4%?

Statistics say the second is more likely.

Anonymous said...

"What matters with three-point defense is not percentage but number of attempts"

Yet you chose to tweet about percentage and ignore attempts...

Hmmm......

Jeff said...

You're again ignoring the fact that Syracuse opponents on average shoot 50% more threes per possession than Wisconsin opponents, and that Wisconsin on average shoots about 33% more than threes per possession than Syracuse does. So coming into the game you expect Wisconsin to shoot about twice as many threes at Syracuse, at a slightly higher percentage.

Instead, they took about three times as many threes at a slightly lower percentage. That's not really all that different.

Jeff said...

I tweeted about percentage, because percentage is where you look for luck. If one team hits 60% of their threes and their opponent hits 15% of their threes then the former team got very lucky. If the former team pulls a big upset, you can put that upset on that lucky shooting. I talk about that all the time in my game recaps.

So I was responding to a lot of people in my twitter feed saying that lucky shooting had put Wisconsin into that game against Syracuse, and pointing out that in fact they hit a lower % of their threes than Syracuse. The hot three-point shooting helped Wisconsin, but it also helped Syracuse. Overall, the outside shooting luck just about evened out.

Anonymous said...

"The hot three-point shooting helped Wisconsin, but it also helped Syracuse. Overall, the outside shooting luck just about evened out."

Nooooooo, Wisconsin's three point shooting at 14-27 was a much greater anomoly than Syracuse shooting 5-9. It's basic math.

Let me ask you this:

Which is a greater outlier, a 40% team shooting 1-2 or a 40% team shooting 50-100?

Jeff said...

And if Wisconsin had shot 50 times as many threes as Syracuse, that analogy would make sense.

The stats said Wisconsin should take twice as many threes as Syracuse, and they should hit them at a higher rate. They instead took three times as many threes at a slightly lower rate. Is there really a big difference between those two scenarios?

Anonymous said...

"The stats said Wisconsin should take twice as many threes as Syracuse, and they should hit them at a higher rate. They instead took three times as many threes at a slightly lower rate. Is there really a big difference between those two scenarios?"

Yes there is. There is a term. Regression towards the mean. The more threes a team takes, the more likely they are to approximate their seasonal average shooting %. If a team takes 9 threes, they are only relative likely to hit near their seasonal average. If a team takes 27 threes, they are MUCH more likely to approximate their seasonal average.

You're predicating Wisconsin's average ("they should hit them at a higher rate") on an assumption that they took relatively equal attempts. That isn't the case.

Jeff said...

Except that all of those percentage that you cite are for a Syracuse team that takes many fewer threes than Wisconsin does, and that allows many more threes than Wisconsin does. So if everything had held exactly to the "mean", Wisconsin would have taken twice as many threes as Syracuse and hit at around a 5% better rate.

Honestly, what is your point here? You must be new to this website if you don't think I know what "regression to the mean" is.

Anonymous said...

Listen, your whole site is based on advanced mathamatics.

If you wanna say that it is equally (or even close to equally) as likely for a 35% team to shoot 5-9 as it is for them to shoot 14-27, fine. Anyone with a highschool math degree knows that's not true, even considering the minescule difference between the teams' 3pt% defense.

Whatever. You're from Wisconsin. You still have Marquette.

Jeff said...

Except that your argument assumes two equal teams playing against each other. We're not. If Wisconsin had taken twice as many threes as Syracuse and hit them at a slightly higher rate it would not have been correct to say that Wisconsin's shooting was further from the mean than Syracuse's was.

By the way, I thought I work for the Longhorn Network. Or I'm an Atlantic Ten hack. Or I set up this whole website just so I can bash Memphis. I can't keep track of what bias I'm supposed to have...

Jeff said...

And by the way, you might notice that I picked Syracuse preseason to win the Big East and to earn a 1 seed. And I picked them to win this game over Wisconsin.

So if I have some anti-Syracuse bias, I think I'm doing it wrong.

Anonymous said...

Answer this ONE question:

What is more likely:

A team averages 34% and shoots 5-9 against a 29.4% defense

OR

A team averages 36% and shoots 14-27 against a 31.5% defense

Jeff said...

How many threes does each team shoot per game? How many do they allow per game?

The answers to those two questions determine the answer to your question.

Anonymous said...

"How many threes does each team shoot per game? How many do they allow per game?

The answers to those two questions determine the answer to your question."

Shots per game and allowances per game are pace-free and irrelavant.

Jeff said...

Shots per possession then.

Again, what is your point here? Are you trying to uncover some secret insidious anti-Syracuse bias in me?

Anonymous said...

"Shots per possession then.

Again, what is your point here? Are you trying to uncover some secret insidious anti-Syracuse bias in me?"

No. I'm trying to show you that if you roll a three-sided die 9 times, you're only relatively unlikely to call the correct number 5 times. But if you roll it 12, 18, 27, or 100 times, you're much less likely to call at a 55% clip.

Jeff said...

That sentence is correct, and totally inapplicable here.


Please acknowledge the fact that if you played the Syracuse/Wisconsin game on a computer, the most likely result would be Wisconsin taking twice as many threes and hitting them at somewhere between a 2-5% higher clip. Taking slightly more threes and hitting a slightly lower percentage? This is really so out of the norm?

I get it, you're a Syracuse fan and you feel like I slighted your team. I have found over the years that the Syracuse fan base is one of the touchier fan bases. But if this is really your biggest complaint with what I've said, you need to find another hobby than clogging up this page with this inane argument.

Anonymous said...

"Taking slightly more threes and hitting a slightly lower percentage? This is really so out of the norm?"

If that's actually what happened, then fine. 200% more threes isn't "slightly more" in most people's books.

"I get it, you're a Syracuse fan and you feel like I slighted your team. I have found over the years that the Syracuse fan base is one of the touchier fan bases. But if this is really your biggest complaint with what I've said, you need to find another hobby than clogging up this page with this inane argument"

Haha. Make it personal. Good job, but you're off base, not that you care. Last refuge. I get it.

ervinsm said...

Just catching up on this convo, but I thought Kpom showed that "defensive 3pt shooting %" is pretty much a myth, and that only taking away attempts is a skill.

Jeff said...

ervinsm, that's correct. I said that in one of these posts. I'm not sure I want to trudge through all of them to find it again, though.

Anonymous said...

"Just catching up on this convo, but I thought Kpom showed that "defensive 3pt shooting %" is pretty much a myth, and that only taking away attempts is a skill."

Exactly. So the more a team shoots, the more likely they are to regress (or progress) to their seasonal average, since the defense has little to do with it. So why, again, is it just as easy to hit 14-27 as a 36% 3pt% team as it is to hit 5-9 as a 34% 3pt% team?

Anonymous said...

"ervinsm, that's correct. I said that in one of these posts. I'm not sure I want to trudge through all of them to find it again, though."

Yet you bring up Wisconsin's 3pt% defense in this thread...

Anonymous said...

Since I can't get a straight answer, anyone feel free to answer:

What's more lucky:

a 35% 3pt team shooting 5-9

OR

a 35% 3pt team shooting 14-29?

Which one is statistically more improbable?

Who cares about 3pt attempts per possession: I'm talking about which team deviated greater from their statistical norm when it comes to made 3ptrs per attempt.

Anonymous said...

* edit 14-27 above.

Jeff said...

I have answered your question as best is possible about a dozen times. This is beyond stupid.

You're obviously going to believe that this whole website is a secret astro-turf operation to smear Syracuse, and nothing I say is going to convince you otherwise. Fine.

Anonymous said...

"You're obviously going to believe that this whole website is a secret astro-turf operation to smear Syracuse, and nothing I say is going to convince you otherwise. Fine."

Wrong again. And you've yet to answer the simple question:

What's more lucky:

a 35% 3pt team shooting 5-9

OR

a 35% 3pt team shooting 14-27?

Which one is statistically more improbable (and 3pt shot attempts per possession has nothing to do with overall %, as KenPom has shown)?

ervinsm said...

Fwiw, heres the EV for each team, but this whole thing is dumb

.36 x 27= 9.72 3's Expected

9.72 EV but made 13s so 3.28 3's above EV for Wiscy aka 9.84 pts above EV



.34 x 9= 3.06

3.06 EV but made 5 so 1.94 3's above EV for Cuse aka 5.82 pts above EV



This whole conversation is kinda silly though. Even more so when we realize that just using a teams raw season 3pt %'s is really skewed and doesnt really tell the whole story (was it contested off the dribble with the shot clock running down, or was it a catch/shoot in the flow 3, as well as was it a 25% 3pt shooter or a guy shooting 40% on the year)

Anonymous said...

Exactly. Which is why saying "Syracuse hit a higher % of 3s than Wisconsin, despite WI being the better 3P% team this yr. So 3s weren't why the game was close" is so dumb.
What if Syracuse shot 1 three pointer and made it? Technically, the same "logic" quoted above would be accurate. But the fact that Wisconsin deviated almost as far from their mean as Syracuse did while shooting THREE TIMES AS MANY SHOTS shows that if anyone got lucky with the three, it was Wisconsin.

Anonymous said...

By the way, Wisconsin was 12.84 points above their expected value. They more than doubled Syracuse's EV on threes. The only reason I'm nitpicking is because I consistantly see this type of nit-picking here, but going the other way.