Big Data isn't just a buzzy concept in the business world these days professional sports teams are, increasingly, applying similar principles to their own operations, analyzing fast-growing quantities of data in order to better understand the game.
A major component of that swelling mass of sports data is positional information particularly in sports like soccer and basketball, where player movement is fluid, and positioning is crucial. There are companies operating in this space, already soccer has Opta, which tracks meaningful events in a game with large teams of analysts manually inputting data in real-time, as well as Prozone, an optical tracking system. The NBA has SportVU, a similar optical system. And the NFL is working to implement a system from Zebra Technologies, which is based on RFID tags in players' shoulder pads.
But these sophisticated analytics products are all proprietary and, presumably, quite expensive, given that the main consumers of their services are professional teams and major media outlets. The average fan doesn't have direct access to their information.
According to Andrew Schechtman-Rook, that's not good enough for the dedicated fan.
"I think sports leagues have long relied on keeping fans relatively in the dark in terms of quantifying team and player performance so that they can control access to how sports are covered in the media," he told Network World. "As a fan, I'd rather get the best analysis out there I don't care if it's from an NFL.com pundit or some guy in his basement."
Schechtman-Rook is an astronomer by profession and a serious New York Jets fan by inclination. He's also heavily into sophisticated analysis of the game of football, and has created the first step in what may be a democratized system of player tracking.
The key, he wrote in a post on his PhDFootball blog, is the NFL's release of game tapes from what's called the all-22 angle a wider-than-normal camera angle used to get a better view of the overall action on the field.
If the camera angles were fixed, it would be easy enough to track players optically, but, sadly, they're not. The abundance of visual references on the field itself, however, lend themselves to the technique Schechtman-Rook has described, which uses them to create an apparently stationary projection of the players moving across the field.
The technology is far from perfect, of course he warns that it's quite hard on computing resources and takes about 10 minutes to process a single play but that's part of why he's taken the open-source approach.
"I'd actually been working on this off-and-on for several months, and part of the reason I posted about it now was the realization that someone else might have the time and the skills to take this much further than I can at the moment," he said.
What's more, his own career path pushed him in the direction of open-source.
"Coming from a scientific background, where transparency is a key part of doing repeatable research, makes me very skeptical of anything proprietary," he said. "How can you trust the analysis if you can't see the raw data?"
Schechtman-Rook's project can be found on Github here.