PSA : Your NCAA March Madness Rules are Garbage. Do This Instead.
On the heels of last year’s fun and successful March Madness Bracket Challenge (“WHERE SKILL PREVAILS!”), we are happy to invite any and all to 2017’s version. Feel free to read the post for this year’s rules, but bear in mind this year’s pool is limited to 250 entrants, so don’t wait:
Register here.
As with most investing topics, our thinking on March Madness bracket rules continues to change. Yes, we still think that traditional bracket rules are a dumpster fire, but our proposed resolution to this matter of international concern continues to evolve.
Why Standard March Madness Rules are Garbage
All the way back in 2014 we identified the core issues endemic to traditional scoring rules. Specifically, standard rules reduce the sample size upon which we judge skill by:
- Encouraging most people to pick the favored team; and,
- Maintaining legacy errors in one’s bracket.
These issues result in diminishing the actual 63-game sample size to an effective sample size of less than 30. And that’s even before we layer on the volatility-enhancing standard of doubling point/win every round.
At the time, we noted that these were pretty straightforward issues to rectify. On the issue of encouraging people to pick more underdogs, we wrote:
…we want the picker to be completely neutral with regards to which team is chosen to win. Ideally, if the rules are set right, half the people in your bracket would choose one team, and half would choose the other, even in the most lopsided games. How do we encourage this distribution of picks? By appropriately rewarding those who correctly predict an unlikely outcome – upsets!
As an extreme example, let’s think about the all-but-overlooked #1 seed versus #16 seed in the first round. In the entire history of the NCAA tournament, a #16 has never defeated a #1. Not ever. Of course this doesn’t mean it’s impossible, simply that it’s highly improbable. In order to entice half of the pool to chose something that has literally never happened before, we must create a powerful incentive to do so. To wit, we want to make the expected returns equal regardless of which team is selected. To see how this might work, imagine that the #1 seed has a 99% chance of winning, meaning the #16 seed has a 1% chance. From the perspective of expected returns, it might make sense to award 99 points to anyone correctly selecting the #16 seed in that matchup and 1 point for anyone correctly selecting the #1 seed.
To make the expected return of each team equal, we simply set the payoff for correctly choosing the favorite equal to the underdog’s chance of winning and the payoff for correctly choosing the underdog equivalent to the favorite’s odds. In the real world, the odds for each team can be backed out by a simple examination of the betting lines. It might not be perfect, but if you believe in the wisdom of crowds, the “sharp money,” or the completely accurate notion that book-makers are profit-seeking enterprises with a vested interest in getting lines “right,” it’s a good enough proxy.
Additionally, in order to eliminate legacy errors, we also suggested choosing all matchups only after they’re known. Again, back in 2014, we wrote:
…we would ideally have every entrant pick every game in the tournament after the matchups were known in each round. In other words, pickers wouldn’t make their 2nd round picks until the entire 1st round was completed, and so on. In this way, every person could pick every game, regardless of whether or not the teams they selected in the previous round advanced.
A Better Way to Score March Madness Brackets
What we failed to consider at the time is that if we fail to eliminate legacy errors, individual matchup parity fails to capture total team parity. And as no reasonable Bracket manager would demand that his pool entrants come back and make fresh picks every round, realistically, team parity is the objective function our rules should create.
As evidence of this, even under our previous scoring methods where we awarded underdogs more point per victory, we still found that the expected total points were still higher for the teams favored to win multiple games.
The following chart from our friend Justin Sibears of Newfound clearly illustrates this issue, albeit with last year’s scoring mechanism.
Source: Flirting With Models, Newfound Research
The good news is that there’s an easy fix for this oversight, too: award points based on the value of a team’s total performance in the entire tournament.
And that’s what we’re doing this year.
ReSolve’s 2017 March Madness Bracket Challenge Rules
Here is the single, simple scoring rule for 2017:
- Teams will be awarded points in proportion to the inverse cumulative odds of their performance in the tournament. The cumulative odds will be calculated as the product of the odds derived from the final lines for individual games from VegasInsider.com, or any other reputable source for live wagering odds. (Note: Any line that is “off” due to overwhelming mismatches will default to 99.5% odds for the favorite, 0.5% odds for the underdog. Any line that is “off” due to a last minute issue – injury or otherwise – will default to the “last” available line.)
What follows is an example of what last year’s scoring might have looked like under this scoring system for Villanova (a highly-seeded team), Wisconsin (a mid-seeded team), and Hampton (a lowly-seeded team), at each round of the tournament.
Villanova | ||
Round | Cumulative Probability | Points |
64 | 0.96 | 1.04 |
32 | 0.77 | 1.30 |
16 | 0.47 | 2.13 |
8 | 0.22 | 4.55 |
4 | 0.13 | 7.69 |
2 | 0.06 | 16.67 |
Wisconsin | ||
Round | Cumulative Probability | Points |
64 | 0.63 | 1.59 |
32 | 0.27 | 3.70 |
16 | 0.11 | 9.09 |
8 | 0.03 | 33.33 |
4 | 0.01 | 100.00 |
2 | 0.005 | 200.00 |
Hampton | ||
Round | Cumulative Probability | Points |
64 | 0.01 | 100.00 |
32 | 0.005 | 200.00 |
16 | 0.005 | 200.00 |
8 | 0.005 | 200.00 |
4 | 0.005 | 200.00 |
2 | 0.005 | 200.00 |
Bear in mind that points are not awarded on a round-by-round basis. Rather, they are awarded on the basis of the final point at which you (the picker) choose a team to win, and they do so. For example, if someone chose Villanova to lose in the round of 16, that bracket would be awarded 1.3 points (the second round being the final point at which the picker had the team winning). Conversely, if they chose Wisconsin to win the championship, they would be awarded 9.09 points (the third round was final point at which the team won).
In reality, Villanova went on to win the entire 2016 tournament, so in this scoring system a picker who had them winning the title would have been awarded 16.67 total points. Wisconsin advanced to the Sweet 16, meaning they would have been capped at 9.09 points for any picker accurately forecasting at least that many wins. And Hampton – poor Hampton – lost in the round of 64, earning zero points. This feels about right, considering the relative values:
- A Villanova championship was roughly equivalent to Wisconsin making it to the Sweet 16/Elite 8;
- Hampton winning a single game would have been equivalent to Wisconsin appearing in the Championship game; and,
- Hampton winning a single game would have been roughly equivalent to 6x Villanova winning the title! (If that sounds disproportionate, consider that #2 seeds – which was Villanova’s 2016 seed – have won 5 titles since 1979, while a #16 has never beaten a #1 seed in the first round, ever.)
If we must endure legacy errors – and we really can’t see any (unoppressively demanding) way around it – then the next best option is to create parity between the total expected values of every team. And the best way we know how to do that is to award points in inverse proportion to a team’s likelihood of advancing to a certain point in the tournament.
So there you have it: if the expected value of every team is the same, how are you going to make your picks? What’s your informational edge? How are you going to identify the signal in a sea of noise?
Let’s see what you’ve got.