A data-driven approach to why I'm picking Iowa to win the NCAA tourney
By: Stephan Teodosescu
A version of this article ran in 2017 and in 2018. My strategy paid off the following year when I picked Virginia to win it all, and they did a year after getting bounced by 16-seed UMBC in the first round.
The pinnacle sporting event of the year is upon us once again. That means we all turn into bracketologists.
The best part about it, of course, is you don't have to be a college basketball expert to win your pool. But it doesn't hurt to try to gain any edge you can. In this post I’m going to explain my methodology and picks for the 2022 NCAA Tournament in trying to win my inter-office March Madness pool.
There are three simple rules I try to stick to when filling out my bracket:
Look for value in picking early round games
Keep your Final Four intact (by eschewing tempting Cinderella stories and picking only high seeds)
Make a “contrarian” pick to win it all
Ben Cohen of the Wall Street Journal put it aptly in his 2018 column: Be chalky early, get crazy late. That is the opposite strategy many try to take. There are so many games ripe for upset in the tournament's early rounds people tend to overindulge.
But the reality is it’s almost impossible to win your bracket pool if you don't pick the national champion correctly. Under standard scoring rules you get one point for every first round matchup you pick right and the points increase in subsequent rounds, culminating in a 36-point bonanza if you pick the title game correctly. As Cohen puts it that should be all the incentive you need to be a contrarian with your choices.
A couple of years ago I came across this great article in Slate that best represents this strategy. The basic premise is you want to find bargain picks in the bracket, not dissimilar from a hedge fund manager’s approach in assessing risk when picking stocks. You want to make boring, safe investment choices that mirror the market for most of your portfolio (bracket), and then dial up the risk by picking an under-the-radar contender for your national champion.
In a 2005 paper, Biostatisticians Jarad Niemi, Bradley Carlin and Jonathan Alexander described a “contrarian” bracket strategy as one that should “favor teams that have a high probability of winning, yet are likely to be underbet by our opponents relative to other teams in the pool.”
A good example of this game theory approach is in 5th Factor's twitter thread below.
This means you'll want to use the wisdom of the crowds to your advantage. ESPN.com's national bracket shows what the rest of the country has picked for every matchup. The goal is to follow the pack in the early-to-middle rounds and pick a winner that may raise a few eyebrows when you talk about it at the water cooler.
My pick: Iowa.
The Hawkeyes are no cinderella story, but aren't exactly a blue blood either. Head coach Fran McCaffery has among the best offenses in the country that helped power Iowa to its first Big Ten Conference title since 2006, and earned the No. 5 seed in the Midwest region this year.
The team is averaging 83.8 points per game, fourth-most nationally and best in the Big Ten. Overall, Iowa is the 13th best team in the land according to Ken Pomeroy's adjusted efficiency margin (AdjEM) metric1. So while they're a strong contender for a title not many folks outside of Des Moines are picking the Hawkeyes for a deep March run2.
Since the tournament expanded to 64 teams in 1985 No. 1 seeds have won 23 national championships (out of 36). Compare that with five for the No. 2 seed, four for the No. 3 seed and just one title for the No. 4 seed. The six, seven and eight seeds also have one championship apiece while a No. 5 has not won since the tourney expanded.
My bet is on Iowa to buck that trend for five seeds.
Picking a 1-seed gives you the best odds to pick the right team. But your friends and coworkers are thinking the same thing. Picking a two, three or even a four-seed comes with a lot of upside as chances are nobody else will pick them; it won't really matter how many games you get right earlier in the bracket since you'll have little-to-no competition at the end.
While it may be tempting to try to pick this year's Loyola-Chicago, George Mason or VCU, keep in mind just 15 of the 144 Final Four teams since 1991 (just over 10 percent) have been seeded seventh or worse. And only UConn (No. 7) has been the non-1 or 2-seed to win it all in the past decade. So stay away from those lower seeds if you can.
Not only do certain seeds win way more than others, but certain conferences are more likely produce winners as well. Don't expect a champion out of, say, the Big West or ASUN because only the best conferences see their teams win championships.
To put some numbers to this the average adjusted efficiency margin for the champion's conference has been +14.6 since 20023. Let's compare that to this season's conferences.
Big 12: +18.0 AdjEM
SEC: +14.2
Big Ten: +13.6
Big East: +12.3
Pac 12: +10.2
ACC: +9.4
The Big 12 is the only conference that is above the 14.6 average. They have three of the top 10 teams in AdjEM, according to kenpom.com. So a win from either Baylor, Texas Tech or Kansas should not come as much of a surprise. Houston, the other team I was heavily considering given the value it represents, plays in the measly American Conference which sports an AdjEM of 7.3. For that reason, and for the fact that Houston just doesn't excite me all that much, I'm picking the Hawkeyes.
It's worth noting though the Big Ten has not won a national title since Michigan State won it all in 2000. Last year was the strongest the conference had ever been with an AdjEM at the end of the season of almost 20. It has seen a drop off this year to a little under 13, but the conference still managed to get nine teams into the dance.
Good luck in your bracket, and for the next three weeks or so, yours truly will be rooting for black and gold. Go Hawkeyes.
______________________
AdjEM is the difference between a team's offensive and defensive efficiency ratings, and represents the number of points a team would be expected to outscore the average Division I team over 100 possessions.
Code to make the above plots can be found on my GitHub page.
List of great bracketology resources: kenpom.com, bracketmatrix.com, ESPN's whom picked whom, fivethirtyeight.com, Luke Benz' basketball central, UIUC's Bracket Odds, Hoop Vision's YouTube video.