In The Beginning

Welcome to TwentyFour, a concept shamelessly inspired by this witch that does political forecasting. That guy has politics & baseball pretty well covered though, so we thought we’d try and delve into the statistics of a sport that we care about a whole lot – the FIRST Robotics Competition. We are also not interested in forecasting for the time being – that’s REALLY hard! We just want to lay out historical trends with the hopes that you can use that data to make better choices. We would like to explain the guiding principles with a quote from Richard Hamming, who determined the atomic bomb being developed by the Manhattan Project would not light the atmosphere on fire.

“The purpose of computing is insight, not numbers.”

If you just want data, there are half a dozen websites and scripts that will list match results, or grab data from the FRCFMS twitter feed, or calculate OPR and DPR. What is sadly lacking is someone willing to sit down and explain what those mean. There is also a recent trend of ranking robots without numbers to back it up, and while that is fun we think there are more useful ways to look at events. We think it is alarmingly easy for people to bring up OPR numbers on their phone and take them as gospel. As they saying goes,

“Garbage in, Garbage/Gospel out.”

We think that one of the best ways to continue to improve the impact of FIRST is to make sure that more competitive (entertaining) matches play out on the field. We hope that by laying out historical trends & explaining metrics we can help teams make smarter choices and ultimately improve the quality of play. Twenty Four teams get to play in eliminations, and we always thought events were more fun when we were “in the hunt”, so we will spend some time trying to quantify what it takes to get there. The first several articles we have planned use the @FRCFMS twitter feed data. We know that this is not a perfect dataset – some regionals are not recorded, some matches get replayed, etc. We have made an honest effort to eliminate obvious issues & the data set is so large that remaining errors do not markedly change the overarching points we are trying to make.  We don’t want to make individual articles too long, but we will shortly post one that lists missing data and why we think its okay to draw conclusions without those matches (hint: sample size!)

We also have the issue of what data to look at. FIRST changes the game every year, which changes our calculus. Since these articles would get rather lengthy if we had to explain games and data, we will try to draw examples from 2010-2012, games recent enough that most of you will have seen them live. If it turns out that 2013’s game closely resembles an earlier year, we will probably draw on it though.

Our first data driven piece will be just how many points and where teams scored points in 2012.

TwentyFour is run by Ian Curtis (writing), Andrew Schreiber (graphs), Jack Sneeringer (data), but if you’ve got something to say you’re welcome to hop on board.  Questions/Comments/Suggestions should go through Ian. (PM me on CD or etotheix at google’s e-mail service)

Extra Credit Reading

The Signal and the Noise: Why Most Predictions Fail but Some Don’t. Nate Silver

FiveThirtyEight. Nate Silver

The Making of the Atomic Bomb. Richard Rhodes