Bracket templates can be downloaded here.

Bracket submissions must be sent here with a subject of “March Madness 2018” by 11:59PM Eastern time on Wednesday March 14, 2018.

On the heels of ReSolve’s past four years of progressive March Madness bracket challenges, we are pleased to take over the reins for the 2018 go-around, and we invite you to participate.

We’ve learned a lot over the past two years of participating ourselves (see this commentary for some tips), and we aim to further their legacy of seeking out bracket rules that isolate the skill of participants and reduce the impact of legacy errors that ruin many brackets scored under traditional rules.

Without further ado, here is the setup:

  1. Prior to the start of the tournament, each team will be assigned a point value equal to 100 divided by the expected number of games that that team is expected to win. This is the team’s points per win (PPW).

    For example, Virginia (1 seed in the South Region) has the highest probability of winning the tournament. Its expected number of wins is about 3.5 games, which translates to a PPW of about 28.6.

    On the other hand, Virginia’s opponent, UMBC (the 16 seed in the South) has an expected number of wins of about 0.02 giving it a PPW of almost 5,000!
  2. At any point in the tournament, teams’ aggregate scores will be their number of wins times their PPW.
  3. Since Newfound is an asset manager, instead of a traditional bracket, each participant will allocate a portfolio of teams. You can give any weight between 0% and 100% to each team, and the weights must sum to 100%. The portfolio score will be the sum of the weights times the aggregate score for each team. For example, if you allocated 10% to Virginia and they won 4 games, you would score 0.1 * 28.6 * 4 = 11.44 points for your Virginia allocation.
  4. Up until now, this has been very similar to ReSolve’s 2016 bracket rules. However, we will multiply the portfolio scores by a scaling factor, S, which is equal to 1.5 – maximum(0.25,C), where C is the portfolio concentration. The concentration is calculated by taking the sum of the squared portfolio allocations. At the extremes: if you equally weighted all 64 teams, C = 1/64 and if you allocated 100% to a single team, C = 1.  This leads to values of S ranging from 0.5 (when C = 1) to 1.25 (when C >=0 .25). In other words, concentrated portfolios will be penalized by up to 50%, and diversified portfolios will be given up to a 25% bonus.

    Why are we doing this?

    One of the issues with these rules without scaling is that someone who chooses a 100% portfolio of the winning team is likely to win. We want to discourage that kind of gambling. With a field of only 64 teams where the seeding has some degree of predictability, there is a decent chance of someone choosing the winner, given enough participants. When constructing a real portfolio of assets, there are many more than 64 securities to choose from, and the consequences of going all in on one (XIV comes to mind) can lead to ruin, given enough time.

We are essentially ignoring the play-in games; those teams have been aggregated into one point total and the play-in game will not count as a win. For example, it does not matter whether Arizona St. or Syracuse wins the play-in game. Your portfolio allocation to the aggregated “team” counts for whoever wins the play-in game.

We have pre-populated a template with the teams, their seeds and regions, their expected number of wins, and their points per win. Download it here, fill out your bracket name and allocations in the green-highlighted cells, and email it back to us with the subject “March Madness 2018”. We will post periodic updates on our blog throughout the tournament.

All entries are due by 11:59PM Eastern time on Wednesday March 14, 2018. Only the final entry for each email address will be logged.

For benchmarks, we will be tracking the following portfolios:

  • Inverse PPW Weighted – This portfolio will earn the same number of points regardless of which teams win (125 because of its 25% diversification bonus). This is akin to the risk free rate.
  • Equally Weighted – Each team will receive a 1/64th
  • Inverse Seed Weighted – Higher seeded team will receive a higher allocation. This portfolio is similar to a momentum tilt.
  • Seed Weighted – Lower seeded team will receive a higher allocation. This portfolio is similar to a value tilt.
  • Front Runner Portfolio – All 1 and 2 seeded teams will receive a 12.5% allocation.

We invite everyone to submit their entries and invite their friends and colleagues. There is no prize, but there is the glory of winning and demonstrating your true selection skill.

May the best quant win!

Nathan is a Portfolio Manager at Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Nathan is responsible for investment research, strategy development, and supporting the portfolio management team. Prior to joining Newfound, he was a chemical engineer at URS, a global engineering firm in the oil, natural gas, and biofuels industry where he was responsible for process simulation development, project economic analysis, and the creation of in-house software. Nathan holds a Master of Science in Computational Finance from Carnegie Mellon University and graduated summa cum laude from Case Western Reserve University with a Bachelor of Science in Chemical Engineering and a minor in Mathematics.