Hey, you, data science startup that has cool technology but is struggling to find a compelling way to demonstrate it. Here’s some free advice: fantasy football. It’s huge business, the National Football League has mountains of statistical data available and, better yet, people might actually pay you for a service that works.
You might think I’m joking, but hear me out. A few weeks ago, a co-worker came to me asking me for any big data tools that would help with his fantasy draft (if he had specific data and specific questions to ask, I told him, he should use Statwing). When I said I was writing this post, someone replied within seconds, “So you’re saying it’s not too late to save my season?”
People love fantasy football — enough to support a whole industry of fantasy football stats services — so it’s a great way to get some attention for something a lot more powerful. If you do it well enough, there could be a business in it.
However, I think the application of proper data science to this field to make predictions more accurate or stats easier to access seems to open the door for a new breed of fantasy statistics. There are probably fewer of those types of services around right now. One that stands out is the Player Comparison tool that SAP provides as a part of the NFL.com fantasy football product.
Here are two more services to add to the list.
SkyPhrase does NLP for the NFL
Even if you don’t care about making money, fantasy football is still a great proving ground for data-based technologies. On Monday, for example, a natural-language search startup called SkyPhrase launched two demo versions of its service — one for Google Analytics and the other for football statistics. The company’s strategy appears to be adding more categories and actually porting its NLP and artificial intelligence technologies on any datasets via API, but football and web analytics make good case studies.
The football tool lets users compare players, ask for specific stats and even set up alerts for when specific things happen on during games. This week, my Green Bay Packers are playing the Detroit Lions. So I searched for the quarterbacks with the most yards against the Lions since 2010. Green Bay’s Aaron Rodgers didn’t make the top 20.
But he has scored a lot of touchdowns against them.
And since 2008, he has absolutely owned them.
And the coolest thing about SkyPhrase it that you can ask it questions via Twitter, too. Send it a question, and it sends back an answer via DM (in like a second, in my case).
Coaches aren’t businesspeople — so sell stats to businesspeople
A couple weeks ago, I spoke with Diane Bloodworth, the founder and CEO of an Atlanta-based sports data startup called Competitive Sports Analysis. Her company started off life trying to sell its analysis to coaches, based on the idea that its unique method of analysis could help coaches figure out what matchups to exploit in any given week. For a quarterback, for example, the company’s software looks at data such as the player’s current stats and his past performance against this week’s opponent, but also data about the quarterback’s offensive line, its past performance against that defensive line, and even the size and strength of relevant players.
But Bloodworth realized that selling software to coaches was a taller order than anticipated. “They’re not like businesspeople,” she said, “I have to think like a coach.” (Which is a diplomatic way of saying coaches often care more about their own expertise than what the data says.)
But you know who is like businesspeople? Businesspeople! Many of them work with data everyday and they embrace it, Bloodworth realized. Plus, many of them play fantasy football.
Fast-forward a couple years to the present, and Competitive Sports Analysis’s primary business is fantasy sports via a product called scoutPRO. Its free version offers preseason draft predictions, as well a variety of weekly rankings about the best players at each position, with scoring projections tailor-made for the specific rules of each about a dozen popular fantasy football platforms. The company has a Georgia Tech Ph.D. candidate on staff whose job is to measure and improve scoutPRO’s algorithms.
For people willing to pay up to $29.99 a season — and many people are — scoutPRO goes even deeper into the stats and will make projections based on a user’s actual roster. The company also sells a version for fantasy baseball, and there’s still an edition designed for coaches, which Bloodworth says is starting to catch on thanks to the company’s success in the fantasy realm.
Of course accuracy is a relative term in a sport where, as the old cliche goes, anything can happen on any given Sunday. A service that could accurately predict those anomalous events where a relatively unknown player breaks out for a career day would be priceless.
And, I’m sorry to say, my GigaOM co-worker who asked about saving his season is probably hopeless even with the smartest analytics around.
Feature image courtesy of Shutterstock user ulegundo.