Bracket Logic: Optimizing Tourney Pool Success

*Data current prior to 2016 tournament

College basketball is a very weird thing. During the regular season, most Americans don’t pay much, if any attention to the sport. However, once March hits, everyone suddenly turns into an expert.   Every wannabe pundit scours the internet, learning every team’s resume to impress their friends, all these hours of work going into the creation of the perfect bracket. However, when push comes to shove and the games are played, these “experts” ultimately end up losing to their grandmother, who picked the teams simply based on team colors.

No one wants to be that guy- the one who studies relentlessly, changing picks up until Thursday morning, and then having his bracket busted in the first few hours of the tournament when that 13 seed doesn’t shock the world. What we all really want is for our brackets to do the talking for us. What most people don’t know is that the key to a successful bracket is extremely simple. In fact, if you’re anything like me, you’ll be annoyed by how simple having a strong bracket really is. This is because the way to maximize your chances of winning your tourney pool goes against the very spirit of March Madness itself: all you have to do is pick the favorites.

You didn’t want to hear that. It’s so mundane. It requires no effort. You hate that guy who turns in the bracket with all the #1 seeds in the Final Four. What’s wrong with him anyway? He might be a pain, and he might not know what he’s doing, but this guy is setting himself up for success. Don’t believe me? Let’s think about it for a minute.

You want to get as many picks correct as possible. In order to do that, you need to pick teams with the greatest chances to advance: aka the favorites. In the early rounds, you want to get every game right so you have more chances to accumulate points in later rounds. The later rounds are worth more points, but it’s hard to determine who will still be in the tournament then. One thing’s for certain: the top seeds are the most likely.

Some people will disagree, with the mindset that if you want a perfect bracket, you have to pick crazy upsets that people would never see coming. This is simply flawed reasoning. Yes, every year there seem to be a couple shocking games. However, by the simple definition of a favorite, it is clear that the most likely outcome consists of all the teams who are supposed to win actually winning. People went nuts picking upsets during Warren Buffett’s billion-dollar bracket challenge and instead overlooked the most effective strategy. Sure, the odds are incredibly high that all the favorites will never win. However, it still remains the likeliest scenario. A #14 may beat a #3 in the first round this year. It would mess up a ton of brackets, but mathematically it wouldn’t be too surprising. The #14 pulls off the upset nearly one time out of six. However, what’s more likely: that all the #3s win their opening matchups, or that you correctly determine the one #14 that advances?

Everyone loves the Cinderella teams, the nobodies that make a tourney run into the Sweet 16 or beyond. However, if you’re going for the best bracket, picking Cinderella schools is a disaster waiting to happen. Let’s say your bracket pool uses the traditional 1-2-4-8-16-32 scoring system. If you pick a #12 to advance to the Sweet 16, you’re making a huge gamble. First, it only happens 16% of the time anyway, meaning you need to be really confident in that pick. Additionally, let’s say they take down the #5 in the opening round, but lose to the #4 in the round of 32. Props to you for that first upset, but it only netted you a point. Everyone who went chalk and had the #4 beating the #5 in the second round missed that 5-12 matchup but earned two points for the 4 seed’s second victory. You came out behind. Picking the favorites is both the safest strategy and the most potent. It shouldn’t even matter which scoring system you use.

On the left below you’ll find a table of every seed, and how often they’ve reached each round up to the Final Four since the tournament expanded to 64 teams in 1985 (31 years * 4 regions = 124 teams of each seed).  On the right is the same data converted into a decimal, where numbers closer to 1 are shaded green and numbers closer to 0 are red.

What the data here shows is that in the opening round, all teams seeded 1-8 are favorites over the 9-16 seeds.  Interestingly enough, the 12 seed fares better in their opening matchup against the 5 seed than the 11 does against the 6, and makes the Sweet 16 more often than the 8s, 9s, and 11s, something I call the mid-major 12 seed effect.  Additionally, the nine seed only makes the Sweet 16 an astonishing 4% of the time, mostly due to coming up against 1 seeds in the round of 32.  Basically, the idea represented here is that all the favorites are the most likely to win, but if you feel the need to go for a Cinderella team, have it be a 10 or a 12.

Now that we know the likelihood of seeds advancing to each round within a region of the bracket, we can use this information to calculate the expected point values of choosing teams to advance to certain stages of the tournament.  For example, in the 1-2-4-8-16-32 scoring system, we can multiply a team’s likelihood of reaching the round of 32 by the one point predicting that game correctly would reward, multiply the likelihood of reaching the Sweet 16 by two, the Elite Eight by four, and the Final Four by eight.  I’ve done just that for four of the most common bracket scoring methods, which are shown below.

While the distribution of points is slightly different in each case, they all share the same basic results.  Choosing the top seeds on average will lead to the accumulation of more points.  Now, keep in mind that all of these formulas are relatively similar in terms of proportion (except for the 1-1-1-1-1-1) and really differentiate themselves in how much they value getting the overall winner correct.  But, they are all normal scoring methods, where the same basic principle should apply.  What do I mean by normal, exactly?  Well, I mean not something like this.

Here’s an example of a really messed up way to score a bracket.  This method (which I’ve actually heard of before, and that scares me) involves you getting the standard 1-2-4-8-16-32 points per round, multiplied by the seed number of the team you correctly picked.  The idea here is to reward picking higher seeds.  However, any pool where the most profitable matchup in a region is picking a #9 in the opening round, and where the seed with the greatest projected yield is the #4 is not a pool you should be playing in.  Funny enough, even with the ridiculous multipliers, it still makes sense to basically go chalk in this and similar scenarios, even if the numbers don’t initially look like it.  If you want to see how that would work, tell me you’re interested in the comment section.

Back to the point.  Picking all the favorites is the optimal strategy.  But that only got us to the Final Four.  Who plays in the National Championship game, and who should be our winner?  Well, going by the same strategy we’ve been using thus far, it would be the best #1 seed on each half of the bracket.  That would be Kansas and North Carolina, who were seeded as the #1 and #2 overall teams by the tournament committee.  The winner would be Kansas.  However, that may not necessarily be the best strategy after all.

Here’s the deal: I maintain that on average, the all-favorites bracket will yield the best result.  However, it isn’t necessarily the bracket that’s going to win your pool.  This is because the winner, Kansas, is being picked by roughly 25% of users on ESPN, and 30% of Yahoo users.  Now, if you’re in a small pool, that’s perfectly fine.  If Kansas wins the tournament (or a team that no one picked does) you have a great chance of winning.  Proceed with the plan.  However, if you’re in a pool with like 100 people, chances are you won’t be the best out of the 25-30 Kansas brackets.  This is when you may want to look towards a different winner.

In a perfect world, teams would be picked to win the tournament proportionally based on their actual chances of winning.  However, since March Madness isn’t the stock market, people tend to choose the best teams far more often than perhaps they should.  This leads to some interesting numbers.  Here are the nine most picked teams in both Yahoo and ESPN pools, compared to their actual odds of winning the tournament, according to Ken Pomeroy.

There’s a lot of numbers here, but what we really need to focus on is the far right column, KenPom/Avg.  This is the percentage of the time a team will win the tournament divided by how often they’re being picked. A number less than one means a team is being picked more often than they are actually going to win, whereas a number over one means the team is being undervalued.  The two most commonly chosen teams, Kansas and Michigan State, are being picked in about half of brackets.  Their KenPom/Avg numbers are around one-half, meaning they are getting picked twice as much as they should.  However, Virginia and Oregon (the other top seeds in Kansas and Michigan State’s regions, funny enough) should actually win the tournament over three times more often than the rate they are getting picked!  This is why in a big pool, these are the teams you want to go with for the best shot at winning, and why I call Virginia the mathematically correct choice as champion this year.

Let’s return to the 100 person pool example.  For the sake of argument, let’s say one of these nine teams wins the tournament, and that the pool winner has the correct champion.  If you pick Kansas, you’ll have the right champion about 15% of the time.  However, out of approximately 28 Kansas brackets, you’re chances of the win are actually very low.  However, if you choose Virginia, you’re right 13% of the time, and you’re one of just four Virginia brackets.  Plus, by picking favorites the rest of the way, you are extremely likely to be the best Virginia.

Here are a few final things to think about before you lock your picks in.  In a pool full of Kansas alums?  Consider having them get upset before the Final Four- you’ll be the only one, which means if Kansas does get bounced earlier than expected, it’s great for you.

When I say pick the favorites, it doesn’t necessarily mean straight 1-8 seeds in the first round.  Interestingly, every #9 seed, plus #10s Syracuse and VCU, are favored by KenPom in the first round.  There are also a few others in later rounds.  If you’re interested, here’s the link for all of KenPom’s Projections.

I didn’t include upset bonuses in my scoring methods simply because there are so many different ways to include them.  If you have a question about your pool, just comment and ask me.  Regardless, if you get an additional point for picking a first round upset, pick all the 9s and 10s.  11s and 12s become coin flips.

Lastly, think long and hard if you actually want to pick all the favorites.  Yes, you might win your pool.  However, you also lose the joy out of rooting for your upsets, and your friends probably won’t appreciate your strategy.  I’ll create the perfect KenPom bracket on Yahoo, and after the tournament, I’ll provide an update on how it did.  My real bracket will be a little more fun, and I advise you to have some fun with your bracket as well.  Who knows, maybe that #13 really will shock the world.