Tournament Stuff (Updated)
Posted: March 23rd, 2009 | Filed under: Other Sports | 33 Comments »
So, I didn’t get a chance last night to really explain the Bill James seeding system. So here goes.
Bill put together a mathematical system that, more or less, figures out what percentage of the time one seed should beat another. According to his system, for instance, a No. 1 seed should beat No. 16 seed a bit more than 99% of the time … and of course a No. 1 seed has never lost to a No. 16 seed in 100 chances.
In fact, the math is pretty easy to figure out this year because the tournament was expanded to 64 teams in 1985, which means that this is the 25th years, which means that there have been exactly 100 first-round games played. Here you can see Bill James math predictions and the actual results of all first round games:
Bill Actual
No. 1 vs. No. 16: 99% 100%
No. 2 vs. No. 15: 94% 96%
No. 3 vs. No. 14: 85% 85%
No. 4 vs. No. 13: 80% 79%
No. 5 vs. No. 12: 74% 68%
No. 6 vs. No. 11: 68% 69%
No. 7 vs. No. 10: 61% 61%
No. 8 vs. No. 9: 54% 46%
Pretty darned accurate. It is fairly interesting to me that the two seed that have really overperformed are the 9th seed and the 12th seed. The 12th seed makes some sense to me because, best I can tell, that’s where the line where the tournament tends to put two kinds of teams:
1. The underachieving team that has been given a second life. This year’s Arizona team fits that tag. The Wildcats in my view DID NOT deserve to make the tournament. That’s because, in my view, they really had a mediocre season. But they are a very good team — they destroyed Kansas at home early in the season — and so given the tournament chance it’s not especially surprising that they are rolling in the tournament. This, to me, would have been like giving the New York Yankees a spot in the playoffs last year. They would not have deserved it … but once they were in, they would have been very dangerous.
2. The overachieving team that is probably better than their seed. This year’s Western Kentucky team fits that tag. The Hilltoppers are a very good team, and they played an Illinois team that was probably overseeded (Illinois was playing without its point guard) and the result was fairly predictable*.
*Though, I admit, I did not predict it. My bracket is a fiasco. I made the mistake I always make — too many upsets. We always hear about all these upsets in college basketball, but it seems to me the tournament is actually becoming much more predictable. Last year, all four No. 1 seeds made the Final Four. And this year, the only meaningful upset was Cleveland State over No. 4 Wake Forest … Wake is the only Top 4 seed that was surprised the first week (No. 5 Purdue also beat No. 4 Washington but that’s not really an upset).
The No. 9 seed beating the No. 8 more than half the time is probably just an indication that once you get that far down in the seeding (The 8 and 9 seeds would represent teams ranked 29th to 36th) there’s really not much to separate the teams, plus the No. 9 seeds have that little extra benefit of being considered “underdogs.”
Anyway, using the Bill system you can figure out, pretty easily — not super easily, but pretty easily — what a team’s chances of making the Final Four are.
For instance:
A No. 1 seed, facing the hardest route available — meaning it would face a No. 16 seed, then a No. 8 seed, followed by a No. 4, and finally the No. 2 seed in the Elite 8 — should make the Final Four about 37% of the time. The No. 1s have actually outperformed that math:
96 No. 1 seeds (entering this year)
14 won championships: 15%
9 finished runner-up: 9%
19 lost in Final Four game: 20%
42 total reached Final Four: 44%
Of course, one of the big reasons that No. 1s have outperformed the math is because they rarely have to face the hardest route available. Take Kansas last year. The Jayhawks beat No. 16 Portland State, then No. 8 UNLV, then No. 12 Villanova, then No. 10 Davidson. A No. 1 seed facing that much easier route should make the Final Four 63% of the time.
So that’s how the thing works. A brilliant reader asked just how hard George Mason’s route was in 2006. Mason was a No. 11 seed, and it had to beat a No. 6, a No. 3, a No. 7 and a No. 1 to get to the Final Four. According to the system, the percentage chance on that is .34%, meaning it was roughly a 294-to-1 shot.
So, now I’ll reprint the Final Four chances for each team still left in the tournament. It’s not that interesting this year because there are only two teams left that are what you would call “surprises” — No. 12 Arizona and No. 5 Purdue. Other than that, the Top 4 seeds are completely intact. We were discussing what this says about the tournament committee: Does it mean they did an incredible job seeding these teams? That was my first reaction, but I don’t know if that’s true. Seems to me there were very few competitive games over the weekend, and very few interesting match-ups. I realize it’s technically not the committee’s job to give us interesting match-ups, but in the larger view, of course, that is precisely their job, right? Nobody wants to watch March Predictability.
Anyway, here again are the percentage chances for each team making the Final Four (using the Bill James seeding system):
Midwest Region
Louisville: 58%
Michigan State: 27%
Kansas: 13%
Arizona: 2%
Comment: Michigan State beat down Kansas earlier in the season … how many times do you think Michigan State coach Tom Izzo will say that it’s “tough to beat a good team twice in the same year?”
West Region
Connecticut: 49%
Memphis: 28%
Missouri: 13%
Purdue: 9%
Comment: The Missouri-Memphis game could be a lot of fun — a couple of athletic teams playing all-out, full-court basketball. Of course, games like that often turn into blowouts too.
East Region
Pittsburgh: 48%
Duke: 28%
Villanova: 13%
Xavier: 10%
Comment: All of the Top 4 seeds are here which should make these games interesting. We’ll see. All the basketball junkies are so high on Pittsburgh, but I didn’t seen anything the first weekend that made me think Pitt is invulnerable. I think Xavier is a tough match-up and, though I dread saying it, I think Duke might be the best team in the region.
South Region
North Carolina: 48%
Oklahoma: 28%
Syracuse: 13%
Gonzaga: 10%
Comment: Again, all of the Top 4 seeds are here. That Syracuse zone is absolutely ferocious, I have no idea how they play the zone better than anyone else. But I would look forward to North Carolina-Oklahoma only to see a Duke player (Jeff Capel) at Oklahoma while a longtime Kansas guy (Roy Williams) coaches North Carolina … it’s just all sorts of baffling.
New Final Four Prediction
Louisville
Connecticut
Duke
North Carolina
Loading ...
Yeah I think you nailed part of it there, really the better the committee does, the more boring the tournament should be. I wonder how comfortable people would be if they were trying to set up interesting matches, like say Kansas versus UNC. I imagine many would say the search for the story was corrupting the process (or something).
I think the toughest road (by seed) any 1 seed ever had was 2002 Maryland, who faced 16 Siena, then 8 Wisconsin, 4 Kentucky, 2 UConn, 1 Kansas, before, alas, running into a 5 seed in Indiana.
Good stuff…
Over/under on Izzo saying “it’s tough to beat a good team twice in the same yearâ€: 12.5 (2.5 times per day).
And it’ll be ironic, since Izzo himself beat the same well-coached team (Wisconsin) four times en route to winning the 2000 championship.
But that’s why we MSU fans love Izzo so much–he exudes the perfect blend of gritty confidence and aw-shucks humility.
Joe,
The thing about Pittsburgh is that they will never beat a team and make you go wow. They just win games. They grind them out in that classic Pittsburgh fashion. Your fatal flaw here is picking Duke. How are they going to beat a team with a strong inside presence? Is Kyle Singler really going to be able to stop a Blair? Probably not. Even so, this team is far too reliant on the 3-pointer, and another fun fact: No NCAA champion has ever shot below 45% during the regular season, Duke is below that mark.
So what’s the point of the Bill James system? If it’s that close to the actuals, why wouldn’t you just use the actuals?
I’ll be impressed if Duke gets by Villanova.
Putting aside seeds and reputations, I thought the two most impressive teams this week were Syracuse and Villanova.
Having picked Pitt to win it all, I hope Duke does get by Nova. Dejuan Blair will rip out Coach K’s heart and eat it in front of him, and that makes for good TV.
Pitt is like Gonzaga though. Sure, they are supposed to be good all the time, but one of these years they might want to, you know, prove it as oppose to under perform. Sure, it’s new players, but it happens to much to be a coincidence.
This tourney has been pretty boring so far. You want the big power-house, blood-spilling, freocious battles in the Sweet 16 and the Elite Eight, but the first two rounds are when you want those 294-to-1 shots to pull one out. Alas, none have.
The only “upset” this weekend was ME. I was upset when I didn’t get to see Gus Johnson’s head explode in person (though Tom Izzo was an eyelash away).
I was upset when they sold out of Dome Dogs before I got to the front of the line (I’m convinced this has something to do with Jason Whitlock).
But I was most upset when I found out that my basketball trek to Minneapolis this weekend denied me the opportunity to see Katy Mixon’s breasts on Eastbound & Down last night.
Hopefully next weekend’s action will do better.
Interesting stuff on the odds of making the final four. However, Joe, I have to take issue with your statement that there were very few competitive games over the weekend. There were very few upsets, but I watched at least part of every game, and many were interesting into the final minute. Siena-Louisville, Gonzaga-WKy, Texas-Duke, and several others were tight games in which the favoured team won. Had a couple of possessions in a couple of games gone differently, we might be talking about another year-of-the-upset. Looking back in 10 years people might not appreciate how near Cal State Northridge came to beating Memphis (the final score isn’t particularly close), but no one who watched the game will forget.
The one thing I’ll say for Pitt is they never seemed to get rattled. Even when ETSU and OK State were holding them close late, they didn’t go into that panicky, “We’re not supposed to be this close” mode that sometimes dooms high seeds in games like that. That could be an important factor in making it as far as the Final Four, although it would be nice if they shot FTs a little better.
Villanova should flatten Duke. Duke was fortunate to get by a mediocre, poorly-coached Texas team. The only way they get by Villanova is if Villanova shoots very poorly and the referees give Duke a lift, as they so often do and as they did, unusually blatantly, in the final minute in the Duke/Texas game Saturday.
Katy Mixon!!!!!!
What was the Marquette coaches deal at the end of the game? Can’t remember the last time a coach just went totally insane and dissed the other coach by running past him after the game. Do Anderson and that guy have a history?
I would far rather get a 10/11/12 seed in tourney than an 8 or 9 if getting to the Sweet 16 is your goal. How many times has the 8/9 seed beaten the #1 seed in 2nd round? Not as often as 10/11/12 seed has won 2nd round game I would expect…
In Tom Verducci’s SI.com article regarding the WBC, he refers to Ryan Braun as “the Brewers third baseman.” Help me out here. Has Braun even played third base since 2007?
The only thing that changed for me in the first 2 rounds is that I’m now a Siena Saints fan. With a break (ok, maybe two breaks), they could have beaten a #1 Louisville team that is listed pretty deep in every expert’s bracket. And having a point guard with only one point on the night, who then hits two three-pointers at the end of two different overtimes — is a perfect example of the vagaries of the tournament.
In a year full of chalk, I wish Siena were still around.
What was the probability of Arizona knocking off three #1 sseds in one tourney, the year they won it all? That has to be one of the greatest runs ever.
John – you do realize that this mediocre, poorly-coached Texas team beat Villanova this year don’t you?
random blog fun:
http://typealyzer.com/
and accompanying article:
http://www.businessweek.com/technology/technology_at_work/archives/2009/03/what_your_blog.html?chan=top+news_top+news+index+-+temp_technology
With regard to #12 seeds, my theory is close to what you said, but using different terms. The #12 seed is a place that the last High Major teams to make the tourney end up (like Arizona this year). Those teams have proven over the course of the year that they can win one or two games against good High Major teams (who are going to be the #4 and #5 seeds that they beat), so it shouldn’t surprise you when they do that in the tourney.
It is also the place that good mid major teams get seeded (not excellent mid major teams like Memphis or Gonzaga, but good ones, like VCU in past years or MVC teams) and those teams are certainly good enough to win one or two games against good High Major teams.
Finally, it is a place where the best Low Major teams (like Western Kentucky) will get seeded. These teams are dangerous because you are never quite sure just how good they are, given the quality of teams they have played in conference for the last couple months.
Should I be surprised at how there is so little West Coast representation? There is Arizona and then Kansas and Missouri are the next closest and they are both closer to the Atlantic than the Pacific ocean.
I’m also confused as to what the point of the James system is. If it were based off of other inputs — team stats, SOS, et cetera — then it would be a great predictive tool supported by a close correlation to actual results. If it’s just based off seeding, why not just use the real numbers, as someone else said?
Marty #20: If you were to plot on a US map (Groucho Marx: “Come over here Mr. Rand McNally and let’s take a look at these maps!”) the location of all Div I men’s basketball programs you would see that a VAST majority of the schools are east of the Mississippi and most of those are near the east coast or at least in the Eastern Time Zone.
Maybe there is some “East Coast bias” because of more major media being located in the east but I don’t think the said bias is overwhelming.
Marty Winn (#20):
Don’t forget Gonzaga, the champion of the West Coast Conference.
Count me among those who thinks the James system is pointless. What makes a #2 seed this year the same as the aggregate #2 seed over the last 30? This would be akin to saying the Rays had a 60% chance to beat the White Sox in the ALDS last season because AL East teams have won 60% of all the playoff series they’ve played against AL Central teams (totally fabricated numbers, but the point stands). James’s system does not take into account the strength of the teams themselves, so it’s fairly useless making any predictions other than “the better seed usually wins”. The folks at Basketball Prospectus, particularly Pomeroy, do excellent work; why not use their numbers instead?
I’m wondering whether the tournament’s increased predictability of the last couple years (last year all no. 1 seeds to the final four for first time; this year all seeds 1-3 still alive after the first weekend for first time) is related to the shrinking pool of at-large mid-majors in the tournament. Only four got in this year. Last year only five or six, I believe.
If the at-large berths are going more often to the also-rans of the power conferences, it makes sense to me that the best teams in those conferences (the 1-3 seeds) could usually handle them in the tournament just as they did in the regular season. Plus there’s not a lot of surprises for them to deal with. Mid-majors sometimes bring a different style of play to a game (and maturity; mid-majors, by their nature, often have senior-dominated lineups).
I think Michael nails it. The non-Big Six conferences got a combined four at large bids, and one of those is Xavier, who no one really thinks of as a big Cinderella. The Committee and the major powers have moved in the recent years to exclude the smaller schools, so they have less representation in the tourney and less chance of upsets. They also gave the mids those dreaded 8 and 9 seeds:
#8 BYU
#9 Siena
#9 Butler
As usual, the mid majors outperformed their seed. There just weren’t many of them to do it. Though Northern Iowa let me down.
Maybe Kansas’ road to the Final 4 was easy by seeding standards last year but I don’t know anyone that would enjoy playind Davidson, Villanova and UNLV in order to get through the meat of the bracket. That is tougher than the seeding suggests.
How did the mid-majors outperform their seeds?
Butler lost to an 8 seed. They performed to their seed.
Siena slightly overperformed by beating an 8 but that’s negated by BYU losing to a 9.
If Xavier loses to Pitt they will have also performed to their seeding.
It seems to me that when the committee does a good job and provides predictable games for the first two rounds, that makes the next two rounds more interesting.
If a lower seed plays out of their mind to make it to the Sweet 16 and then gets blown out, that’s just switching the timing.
Bigtime college sport is already way overemphasized. Consider football salaries (the above-board ones to the coaches), recruiting and eligibility scandals, too-long seasons, the amount of maiming (Doonesbury’s wounded vet will feel right at home hanging around football practice because so many of the guys also have TBI). And then we get to the “idea” that a college football playoff would “prove” something, just like March Madness supposedly does.
But our comments here are about “the vagaries of the tournament” and interesting upsets. What does an upset “prove” except nothing — that on any given day of the week, one of these teams might have a hot streak or a streak of bad luck. No matter the result, we’re no closer to knowing which is truly the better or best team — not unless they spend the entire spring semester imitating the NBA playoffs.
With football, the fraud (that we had identified a “true” champion) would be greater, because of the funny way the pigskin bounces. I’m for fewer games and less money spent on this stuff — more time and attention and money spent fixing our REAL problems. Bah. Humbug.
[...] [...]
Of course, the way to get the best matchups and the only FAIR way is to not seed tournaments in the first place. Why are tournaments seeded? To give the most possible advantage to the “best teams”, but what’s the point of playing a tournament if you already know which teams are best? Throw the 65 teams into a lottery and set the matchups randomly. It might make Bill James’ task as oddsmaker a bit more difficult, but it would provoke a lot more offseason commentary. The teams with the most expensive programs might decide not to chance playing in unseeded tournaments if they thought they might play Uconn or North Carolina in round 1.
[...] Release: All-American Teams | College Chalktalk…SLAM ONLINE | » Breaking Down the Tourney…Tournament Stuff (Updated) » Joe Posnanski…Last Minute Bracket Guide/ Tourney Preview | IWS Sports Blog…The Starting Five Sweet Sixteen [...]
“It make sense that 12 seeds are underrated”, said the Yale graduate as his stomach rumbled – longing for cheese.