Hall of Fame
Guru's Market Madness Contest
In past years, I was frustrated by the inability to find detailed historical tournament data upon which to perform analysis. Sure, you could find a list of the tournament winners, and even final four data was not hard to find. But drilling down to the earlier rounds was virtually impossible, at least for all but the most recent years. So, I decided to do my own research.
What you'll find here are tables of data and analysis that covers the tournaments from 1985-2017. These are the 33 years during which the field has consisted of 64 (or 65 or 68) teams. While that's not necessarily a lot of data upon which to develop credible averages, the resulting records do seem to make intuitive sense.
First, let's take a historical look at how teams would have done using this year's contest scoring. In 33 years, we have data for 132 #1 seeds, 132 #2's, etc. Following are some statistics grouped by seed:
The table points out an important statistical feature of a single elimination tournament like this. The distribution of results is not symmetrical. This is because half of the teams in any year lose their first game, and are done. Three-fourths of all teams fail to win more than one game. You can see the impact of this by comparing the median score with the average score. Median scores are all pretty low (some are even negative), while averages tend to look more similar to the 75% percentile scores. (Actually, the averages aren't quite that good, but hopefully, you see the point I'm making.) So the typical distribution of results includes lots of lousy scores, a few reasonable scores, and (depending on the seed) a handful of great scores.
The 2000 Contest introduced the availability of "seed basket" units. A seed basket contains all four teams with the same seed. For example, in 2017, the #1 seed basket would have included the combined results of Kansas, Gonzaga, N. Carolina, and Villanova. The following table shows the range of results that seed baskets would have experienced over the past 33 years.
If you compare these maximum potential values to the actual best results by seed (above), you'll note that no team has yet produced the maximum. The best #1 seed result fell only G$1 short of the best possible, and Villanova's 1985 win as a #8 seed was only G$2 shy of the maximum potential for a #8 seed. (In the 3rd round, 'Nova faced the #5 seed instead of the #4, and in the first round of the final four they faced a #2 seed instead of a #1.) Arizona had a tough path to win in 1997, beating 3 #1 seeds along the way, but they faced a #12 seed in the second round, and a #10 seed in the fourth round, thereby falling G$15 short of the maximum potential for a #4 seed.
RotoGuru is produced by Dave Hall (a.k.a. the Guru), an avid fantasy sports player. He is not employed by any of the fantasy sports games discussed within this site, and all opinions expressed are solely his own. Questions or comments are welcome, and should be emailed toGuru<firstname.lastname@example.org>.
© Copyright 1998-2018 by Uncommon Cents, LLC. All rights reserved.