Former President Barack Obama is a big sports fan. In fact, for every year while in office, he filled—and made public—his brackets for the NCAA Tournament. And it’s a tradition that he’s kept up since leaving the White House.

Word is that he was pretty good with his picks. But how good? Is Obama good at predicting upsets? What is his favorite school? And when did he fill out his best bracket ever?

Answering bracketology questions is always a bit tricky because methodology matters. How things are valued matters. A lot. So before we start analyzing all 630 of Obama’s picks, laying out those assumptions is important. Let’s assume, for now, that we are using the most common scoring system out there: 1-2-4-8-16-32, as in 1 point per correct pick in the first round and 32 for correctly predicting the national champion. This methodology is used across the board, including by major outlets such as Fox Sports and CBS — ESPN also uses the same proportions but scaled to ten (10 through 320). We will call this one the standard system.

In the last decade, three years were clearly the best ones for the former President: 2009 and 2012 in his first term, and 2017, the first year after the end of his presidency. Obama’s second term was pretty much a slump and last year, well, did not look good at all. He had an all-time low of 35 correct picks, well below the Obama average of 39.4.

In 2012, the former commander-in-chief had 44 correct picks, meaning he chose winners correctly just about two-thirds of the time. Correct picks, however, do not necessarily result in high bracket scores. 2012 is only his third best bracket score. 2011, with 43, holds the worst point-per-game rate among the ten brackets.

Obama peaked at 123 points in his very first year as president, in 2009, propelled by an impressive 28 out of 32 points in the second round and predicting North Carolina’s national title.

And that is the reason why methodologies matter. The more of the last four you pick in a system like the standard, which values games played later in the tournament more, the more success you will have, even with a mediocre performance in early rounds.

In fact, Obama’s 2009 bracket was so good that it wins in almost every point system that has been used. But 2012, a year where the former president only got half the Final Four correct, is a close runner-up as the weight of games is more balanced. To be sure, while Obama’s predictions in 2009 were impressive, that was also a year of high-seeded teams going deep into the tournament. It was a pretty predictable year.

Rewarding Madness

One of the most interesting features of the NCAA tournament are the upsets. What if we introduced that madness into Obama’s brackets?

First, let me make the case for it. In 2010, President Obama trusted top-seeded teams like Kansas, Kansas State, Villanova, and Kentucky to reach the Final Four. That would be two #1 and two #2 seeds. None of those teams made it there. Similarly, in 2018 the bracket was busted open by the first-ever 1-16 upset of Virginia against UMBC in a year when his bet was on Virginia reaching the final against Michigan State (who lost in the second round). That doomed any chances he had to make a splash in those tournaments by the standard scoring, even though he correctly guessed some unlikely outcomes. Which is why, since an upset like that can cost one dearly, predicting those correctly should be more highly rewarded.

One simple way of identifying upsets is comparing how the teams were seeded by the NCAA selection committee. Ultimately, these ranks will serve as a relative measure of strength between teams (for more advanced ways of doing this, check out the ELO methodology used by FiveThirtyEight). An upset would happen when the lower-seeded team wins the matchup.

The way I decided to reward upset pickers was to give an extra point based on the difference of the teams’ seeding and the tournament round when the game is being played. After testing multiple complicated calculations and weights, I settled on one simple formula:

UPSET POINTS = POINTS WORTH * |(HIGHER SEED – LOWER SEED)|

For instance, in the aforementioned Virginia-UMBC upset, a correct pick in that matchup would yield an extra 15 points. Had that matchup occurred in the regional finals – the first opportunity after the first round where a #16 seed can meet a #1 seed – such an upset would result in 120 extra points.

Factoring that in, 2012 gets a 42 percent boost to Obama’s brackets, giving him 140 points in total. The most impressive jump would have happened to his 2010 score, when Obama predicted 7 of 15 upsets in the tournament. That would have boosted him by an extra 28 points, or a 47.5 percent increase over the standard system.

Other upset weighing calculations more than double 2010’s total points. For the sake of sanity, let’s not get into them.

Is Obama a bracket genius?

We do not have complete data to compare every one of the former President’s brackets to average performance. But between 2011 and 2018, the data provided by the NCAA shows that Obama is slightly better than the average under the standard points system. The exceptions are 2015 and 2018, when Obama’s bets did not pan out, leaving him in the negative when compared to the average.

But in 2017, Barack Obama outperformed the average of NCAA pick’em players by 70 percent. Maybe that was actually his best bracket ever?

Unfortunately, there’s no way to ascertain how good 2009 was against the field of players. With all the high-seeds reaching regional finals, a high field average would be expected.

Picking favorites

Coincidentally, those years ended up with the president correctly picking his favorite team, the Tar Heels, as champions. In fact, North Carolina is the team the president bet most points on. But even with NC’s two titles helping his cause, Obama actually tends to overbet on North Carolina. That is also true for powerhouse programs like Kansas and Michigan State, who have not won since Obama’s 2008 election.

Surprisingly, one of the most successful teams of the 2000s, Duke, received substantially less confidence from Obama. In fact, Coach K’s team was undervalued by the president, as was Villanova.

But the biggest discrepancy between Obama’s trust and actual results is seen with UConn. The team went on to win the national title in 2011, but never got past the third round in the former president’s brackets.

Kentucky, also a national champion in the period under consideration, and Louisville, who had their 2013 title vacated, fall in the sweet spot where Barack Obama bet on them about the same number of points as they amassed since 2009.

The 2019 NCAA Tournament is upon us and, well, I’m personally excited about a bunch of things, including the chance for the #13 Northeastern Huskies to upset the #4 Kansas Jayhawks.

As for Obama’s 2019 bracket, we are still waiting, Mr. President.