Selection Sunday: Is your March Madness prediction algorithm ready? | Hacker News Reader

Selection Sunday: Is your March Madness prediction algorithm ready?

blog.smellthedata.com

10 pointsdanger13 comments

If you're interested in this kind of stuff, Ken Pomeroy (kenpom.com) runs a fantastic basketball analytics site. He's a big proponent of what are called "tempo-free stats", which aim to filter out issues of playing speed from scoring (a team that plays quickly will score a lot of points, nearly independently of whether or not they are winning). Tempo-free stats instead count possessions; one interesting statistic is that the average team this year in college basketball produced 1.01 points per possession - such a tidy figure to emerge from the chaos.

In terms of predictions, one of the most interesting teams this year is Kansas (http://kenpom.com/team.php?team=Kansas). They've only lost twice but have a large number of narrow home wins. Depending on how your algorithm treats those wins they either look like a team that will struggle to reach the sweet 16 or like a potential national champion.

One other issue that comes up is "garbage time". When a game isn't close (say in the last quarter of a blow-out), the stats are basically meaningless. Does Pomeroy have good ideas about how to deal with that?

Funny, I did tempo-free stats when I was about 10 years old -- decades ago. Although I didn't have access to actual possessions, I tried to base it, best off I could off of field goals attempted for individuals and at the team level, and do various extrapolations.

Seems like it is still a fun area, with much better data now. I just need the time.

Why can't people use any dataset they want? I think the rule should be, "Use any data you like, but you must submit any data used for use by the rest of the field".

This was the intention. Around a month ago we started asking what data people would like to use. We incorporated some of that feedback to decide what data to use for this year.

If you have other suggestions, please let us know, and we'll add it for next year (if possible). The only thing we're trying to avoid is somebody coming in with a lot of data at the last minute, beyond the point when anybody else can realistically get it incorporated into their model.

Actually that makes sense. I didn't realize that you did the request for data earlier. I may have to do this next year, if you're still doing it.

Yeah, unfortunately the "getting the word out" could have been done better.

See my unrefactored-ruby-from-two-years-ago solution. https://github.com/listrophy/lazy_ncaa

It's pitifully simple... it just uses the historical probability of seed A beating seed B in round C. Going into the Elite 8 last year, it produced a bracket in the top 800 on ESPN.com. Of course, the other three brackets it produced (and I posted) failed miserably.

What does last years algorithm look like on this years tournament.

I haven't run it yet. It should be done in time to enter the contest, though (but it will just be a baseline--i.e., not eligible to win prizes).

Oh boy, basketball. The only thing less interesting than Gruber's baseball related posts, other than Golf on TV.

I just can't get into watching sports - I love playing them (soccer and ultimate frisbee on the weekends) but watching them seems pointless for some reason. My wife watches more sports than I do, and she watches figure skating and track & field, which comes on like 4 times a year...

Anyone else feel this way?

I had to reply because my wife watches more sports than I do as well. When the in-laws get together for things like the Super Bowl they let me hide in the basement to work. I'm definitely no good at actually playing sports, but I'd much rather do that than watch someone else play them.

Reply on news.ycombinator.com