A Math Geek s Guide to Bettering Your Bracket

If you’re looking for tips for filling out your NCAA bracket, you can flip over to ESPN, or you can turn to a Ph.D. at the University of Cincinnati who has spent decades working out the math behind choosing winners and losers.

Michael Magazine, a UC Carl H. Lindner College of Business professor and Ohio Eminent Scholar in business analytics, has made it his passion to understand the science behind sports numbers.

• Jump to his Five Quick Office Pool Tips

“You have a lot of people who are sports geeks, and a lot of people who are math geeks,” he says. “And you have some people who are both. I fall into that category.”

While plenty of folks take sports and numbers to heart (sports betting is a multi-billion dollar industry) a handful of high thinkers simply enjoy diving into digits for the sport of it.

For example, Magazine and a half-dozen of some of the brightest math minds in the country — hailing from such institutions as MIT, Georgia Tech and City University of New York — have had a 30-year running competition as to whose probability models will most accurately forecast the NCAA tourney champ. They also pit their wits the rest of the year, too, by predicting World Series, Super Bowl and NBA Finals champs.

In their homegrown pool, however, each player is allotted 1,000 points to cast prior to the start of the tournament in a sealed bid auction for teams. Whoever bids the highest for any given team “owns” that team. Magazine, who does better in his brainiac pool than the typical office pools he enters, can lay bragging rights to the latest NFL season in which he successfully predicted both Baltimore and San Francisco would make it to the Super Bowl.

Unlike the average office pool, this group, which he affectionately named “The Suckers,” takes more than just a completed bracket to gain entry. In this fraternity of sorts, a Ph.D. is the prerequisite and playing along requires both a love of sports and an even stronger fondness for figuring.

But even with all the formulas that go into calculating the probabilities, getting NCAA games right as often as 60 percent of the time is considered outstanding. And getting all the games right when it comes to the NCAA tourney is virtually impossible.

“Nobody in history has ever gotten a perfect bracket” says Magazine. “The number of possible realizations for the NCAA tournament is more than 9 billion billion. The number of combinations has 18 zeroes. It is staggering.”

When Magazine isn’t mind sparring with his sports-minded contemporaries around the country or rubbing elbows with the likes of NBA owner Mark Cuban at the annual MIT Sports Analytics Conference, he’s teaching and researching complex algorithms to streamline such things as

health care delivery and supply chain management.

He also teaches a “Sports by the Numbers” course in which students weigh common in-game scenarios: Should a coach go for the yards or punt on fourth down? And what’s the best approach for a basketball team that finds itself down by two with seconds remaining? Go for the tie or shoot a three? By the way, Magazine says teams are more likely to win the game if they go for the three than taking their chances in overtime.

This year he introduced a three-Saturday class leading up to March Madness called “Bracketology.” In that course, students spend three eight-hour Saturday classes learning not just the history of the NCAA tournament but also how to run probabilities and game simulations.

Magazine co-teaches the course with former student Paul Bessire, who completed his masters in 2005 in quantitative analysis. Bessire, a Lindner Honors-PLUS graduate, was once responsible for the math behind the historical simulation sports games at WhatIfSports.com. Today, he owns

PredictionMachine.com

, which draws upon his advanced forecasting software to calculate likely sports outcomes. This time of year, Bessire is commonly a guest on national sports radio programs and is a regular

blogger at the Huffington Post

.

Bessire’s algorithms take into account the most important statistics behind all 68 teams in the tournament then simulates the entire bracket 50,000 times to predict the most likely winner. Who takes it all this year? According to Bessire's calculations, it will be Indiana. And he's successfully predicted the winner seven times in the last nine years.

Still, Magazine cautions, while predictions can be made “you are never going to know for sure who is going to win.”

Five Quick Office Pool Tips

1. Ignore the rankings. Instead, compare such stats as turnover percentage, field goal percentage, offensive rebounding percentage and free throw rate (that is, a team’s number of field goals divided by their number of free throws attempted). The more times a team heads to the line, the better.
2. Don’t force upsets. It’s more important to keep teams alive in later rounds than to be the leader after day one. So, only take chances with close matchups.
3. Don’t pay attention to seeds. Seeding is more about deciding which team is “most deserving” than which is actually favored to win. Dig into each head-to-head matchup to determine which team is actually the favorite.
4. Path is more crucial than seeding. Regional difficulty and the journey a team will have through the bracket actually matters more than the seeding when it comes to choosing your late-round favorites.
5. Ignore your allegiances. Try to put subjective information aside (including team colors and favorite mascots) and look purely at teams objectively.
   

Video:

UC graduate Paul Bessire shares his predictions with CBS Sports