Print and Go Back ESPN.com: Stats & Info [Print without images]

Tuesday, March 18, 2014
Filling a bracket, with BPI's insight

By Alok Pattani, ESPN Stats & Info


Ralph Freso/Getty ImagesBehind Nick Johnson, Arizona is No. 1 in BPI and the most likely team to win the NCAA championship.
While ESPN’s Basketball Power Index wasn’t designed to be purely predictive, there has been interest in using BPI to help with picking NCAA Tournament brackets. It’s one thing to just look at the BPI rankings and pick the higher-ranked team to win each matchup. If you do that this year, you end up with a pretty chalky bracket with all four No. 1 seeds making the Final Four and Arizona over Florida in the title game.

A more nuanced way to look at things, however, is to determine the percentage chance of each team getting to each round. This can help not only by taking into account the effects of opponent strength round-by-round (beyond just “better BPI”), but it also helps illuminate what might be considered smart upset picks and undervalued or overvalued teams.

With so much randomness in a single-elimination tournament, it’s important to know how certain (or uncertain) you should be as you fill out each line of your bracket.

With the BPI game prediction calculation we’ve used throughout the year, each possible tournament matchup can be projected using each team’s current BPI. After calculating the individual matchup probabilities and accounting for every possible way the tournament could unfold, the probability of each team reaching any given round can be found.

Below are the 12 teams with at least a 10 percent chance of making the Final Four this year, along with their percentage chances of going beyond that.

The chart shows that although the most likely Final Four according to BPI consists of the four No. 1 seeds, the likelihood of all four of them getting to North Texas is 1.3 percent.

The most likely scenario is that only one of the No. 1 seeds makes it, as has been the case in each of the last two seasons.

The East is the most “open” region with each of the top four seeds having between a 14 percent and 27 percent chance of making the Final Four. The West is the most lopsided, with Arizona’s 47 percent chance much better than 2-seed Wisconsin’s 13 percent.

Although Arizona is the “BPI favorite” to win the title, keep in mind that there is nearly a 4-in-5 chance that another team will win it. The fact that the top teams are a bit closer together this year and the randomness of a single-elimination tournament make the field a much better bet than any single team to make the Final Four, and certainly to win the championship.

Finding value
If you compare these values to what the public is picking in ESPN.com’s Tournament Challenge, you can find “value” picks or teams to avoid. For example:

• Villanova has a 24 percent chance of making it to the Final Four. Less than 10 percent of the public is picking the Wildcats to get there, which might make them an undervalued selection.

• Conversely, Michigan State has a 15 percent chance of making the Final Four. Even if Michigan State’s BPI rank of ninth understates the team's true quality going forward, it’s unlikely the Spartans are as high as 44 percent likely to make it to North Texas, which is the rate at which the public is picking them.

• Florida’s chances of making it to the Sweet 16 and beyond are significantly overestimated by the public, with the Gators being picked 94 percent of the time to reach the South Regional in Memphis, Tenn., 62 percent of the time to reach North Texas and 28 percent of the time to win it all.

To get to the Final Four, the Gators could face an underrated Pittsburgh team in the Round of 32 and might have to beat a strong Kansas team for the second time this season in the Elite Eight, so BPI doesn’t see their advancement to be as likely as fans do.

Keep in mind that this system is based on Selection Sunday BPI and the structure of the bracket itself. It has no additional factors included and is not fit to past years’ data or anything like that as would be done in a traditional predictive model.

Given this and the fact BPI isn’t designed to predict the tournament, this fairly simple projection has performed pretty well the past couple of years.