Friday, March 15, 2013

Who?s No. 1? Who Cares.

Indiana University forward Cody Zeller (40) holds the basketball while being defended by Butler University center Andrew Smith during the first half of their NCAA basketball game in Indianapolis December 15, 2012.

Indiana University forward Cody Zeller (40) defended by Butler University center Andrew Smith

Photo by Brent Smith / Reuters

As we count down the days to Selection Sunday, speculation will intensify about the bracket for this year?s NCAA Tournament. Since the beginning of the regular season, and with increasing fervor over the last few months, dozens of sites have assessed and reassessed which teams will get the four prestigious No. 1 seeds. All year long, the top of the field has been incredibly scrambled?ESPN.com?s Bracketology has tabbed eight teams (Duke, Indiana, Louisville, Gonzaga, Kansas, Georgetown, Miami, and Florida) as potential top seeds in just the last few weeks.

Here?s some advice for bracketologists and fans alike: In terms of determining a champion, it doesn?t really matter which of the top teams gets slotted where. Every March when the field is announced, I compute the chances of all 68 teams advancing to each round of the tournament based on my own rating system. While the vagaries of seeding and bracketing can dramatically alter the chances of, for instance, a No. 12 seed making it to the Sweet 16, I?ve always been amazed at how little bracketing?and especially seeding?affects the chances of any team winning the title.

The order of all these teams in my ratings (or whatever rating system you prefer) will be very similar to the order of their chances of winning the event. To whatever extent those odds differ, the collective strength of the teams in a particular region will have a larger influence than any seeding irregularities.

Consider last season?s title-winning Kentucky team. It didn?t take a basketball expert to realize that Anthony Davis, Michael Kidd-Gilchrist, and crew should be a top seed, and Las Vegas made them a clear favorite to win the championship. We can get a rough idea of how much their seeding mattered by using Bill James? log5 model (in simplest terms, a statistical technique to estimate the odds that a team with Strength Rating A will win a game against an opponent with Strength Rating B), which I?ve relied on for years to make my pre-tournament predictions. This model can give us a crude estimate of Kentucky?s chances of winning a title under various scenarios.

Let?s assume the Wildcats had a 30 percent chance of winning the title before the tournament began?a reasonable guess given the odds that casinos were offering. Given that scenario, Kentucky could have been mis-seeded very badly without seeing their chances suffer significantly. To test this, I dropped UK to a No. 3 seed in the South Region, moving Duke and Baylor to the 1 and 2 line, respectively. This dropped the Wildcats? title chances by just a half of a percentage point. That seeding mistake would have been the biggest blunder of all-time, and yet it wouldn?t have affected Kentucky?s ultimate objective of winning six consecutive games.

What?s more important than seed is the difficulty of the region a team is placed in. Moving Kentucky to the West, where higher-seeded teams were relatively weak, would have caused their chances to spike to 37.5 percent, while playing in the East region, where a very tough No. 2 seed in Ohio State was lurking, would have sent Kentucky?s chances tumbling to 22.8 percent.

Why doesn?t seeding significantly influence title chances? A team?s chances of winning the tournament will be restricted most by its potential matchups against the great teams in the field. Most of the time the great teams will be found in the Final Four?if they?re even there at all?and the chance of playing one of them doesn?t differ much whether a team is a No. 1 seed or a 3. So before the tournament starts, the chance of any team facing a great one is going to be pretty similar. (One exception is when those great squads are lurking in that team?s own region, as was the case in our theoretical example of Kentucky and Ohio State in 2012.)*

This season, much of the seeding debate centers around Gonzaga. Ranked No. 1 in the AP poll, most bracketologists project the Bulldogs to get a 1 seed. But given the team?s lowly strength of schedule (72nd according to the RPI formula, the lowest of any candidate for a top seed), it?s possible Gonzaga could fall off the top line. Indeed, because the West Coast Conference offers few games against top competition, the Zags? seed tends to be more difficult to predict. In 2002, Gonzaga was ranked sixth by the AP but received a No. 6 seed, and in 2006 they got a No. 3 seed despite being ranked fifth.

One thing we can be very confident about is that Gonzaga will be placed in the West region. Because western conferences typically yield a disproportionately small number of elite teams, they have a better chance of getting geographic preference. Using Wednesday?s projected bracket from ESPN?s Joe Lunardi, we can look at how seeding will affect the Zags? chance at a title.

As a No. 1 in the West, Gonzaga would have a 10.6 percent chance of winning the title using Lunardi?s bracket. Swap them with the No. 2 in the West, Miami, and their chances of winning it all would actually increase to 11.6 percent. In their worst-case scenario, a No. 3 seed, the Zags? chances would be at 10.8 percent. As with the case of the 2012 Kentucky Wildcats, Gonzaga?s chances are less sensitive to their own seed than to the other teams in their region. If Indiana were placed in the West instead of Miami?which at this point is extremely unlikely?the Zags? chances with the No. 3 seed would drop to 8.7 percent.

Still, the effect of seeding and draw is rather small when you?re looking at the overall chance of winning the NCAA Tournament. This can be further illustrated by estimating the title chances for each team in Lunardi?s bracket. Some teams? odds improve based on their seeding and draw, while others take a slight hit, but the order of each team?s chances largely follow whatever assumptions you make for team strength.

Plugging in data from my model, the top four teams?Florida, Louisville, Indiana, and Gonzaga?also have the best chances, despite Florida not being a No. 1 seed in Lunardi?s projection. This isn?t just an artifact of my model. If you believe Duke is the best team in the country, then they would almost surely be the favorite to win the title regardless of what the real bracket looks like when it is revealed on Sunday. They?re the fifth-best team in my model, and have the fifth-best chance of winning the tournament based on Lunardi?s bracket.

The worst mistake the basketball committee can make is to leave a deserving team out of the field. Once a team makes it in, seeding mistakes matter most in the bottom half of the bracket. If a Cinderella hopes to make the Sweet 16, it needs to be matched up against teams it can beat. But fans of Duke, Indiana, and Gonzaga, whose objective is to win it all, shouldn?t care about the little number next to their team in the bracket. The debate over who should be No. 1 is a fun intellectual exercise, but when it comes to cutting down the nets in April, it hardly matters at all.

Update, 3:30 p.m.: This story has been updated to clarify that a potential regional pairing of Kentucky and Ohio State was a hypothetical, not an actual matchup from the 2012 NCAA Tournament. (Return to the revised sentence.)

Source: http://feeds.slate.com/click.phdo?i=d29d2008db9a97a2abb73b3aaa89839a

NBC Olympics NBC Olympics schedule 2012 Olympics Chad Everett London Olympics Kristen Stewart Rupert Sanders Photos 2016 Olympics

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.