“Structured data versus unstructured data” – that doesn’t seem like the kind of phrase uttered by fantasy football players discussing strategy over a couple of beers, but perhaps one day soon, it might be.
It’s now ten years since IBM’s celebrated computer system, Watson, competed on Jeopardy!, demolishing the human champions Brad Rutter and Ken Jennings and landing the $1 million top prize, albeit with a few strange hiccups along the way. It was heralded as a big moment for machine learning, and generated the kind of buzz IBM desired.
As most fantasy football competitors are aware, Watson also entered the game a few years ago. There was a big push with new features for the ESPN-IBM fantasy football app. It’s fair to say that it’s been a mixed bag. While data scientists have sung the praises of Watson’s fantasy football predictions, players aren’t so sure. Go to forums on Reddit, and you’ll mostly see a bunch of bemused commentators. While there is some praise for the “Boom or Bust” feature, many players can’t see the allure of Watson.
We aren’t going to discuss the pros and cons of Watson, as such, here. The system is not perfect, including in areas outside sports predictions. But does anyone really believe that it, or systems like it, won’t eventually conquer the world of fantasy sports? The Jeopardy! appearance was a decade ago: where will we be in another decade? Computers capable of crunching vast quantities of data will be able to turn that into unbeatable fantasy football strategies – you can guarantee it.
Machine learning will transform many industries
Fantasy football won’t be alone either. With the rise of sports betting across America, thanks to the 2018 decision by the Supreme Court, computers and analysts might also be able to beat ‘the system’ for profit. Sportsbooks, of course, use their own algorithms and data scientists, allowing them to confidently update live betting odds in real-time. But we can foresee a scenario where it is computer versus computer, rather than intuitive punter versus bookmaker. Beyond sports, IBM has plans for Watson to enter the arena of law and medicine.
For some, that might be a good thing. Even now, players can access data sets to help them get an edge. But you have to wonder where it will lead. Winning is fun. And the means can justify the end, in a sense. Yet, there is a loss of – let’s call it, romance – that is the reason we love the game in the first place. That feeling of triumph when your gut has told you Alex Smith will have a big comeback season for Washington, and it actually happens – that’s the romance of the game. The feeling is different when a machine tells you it is going to happen.
The fun might disappear from fantasy sports
Going back to what we mentioned earlier: structured data versus unstructured data. The former is easy-to-understand, quantifiable information – box scores, player touchdowns, win-loss records, and so on. The latter, unstructured data, is the unquantifiable information – emotions, expert tips, journalist reports, countless fantasy football blogs, etc. Machines like Watson are capable of reading thousands of articles on these topics and turn what it reads into usable data.
The machine might learn, within seconds, that Tom Brady’s offensive performance level drops ever so slightly when his wife is in the news headlines, or his family is abroad, or when the weather is two degrees below average in Foxboro. That’s what IBM claims is the difference-maker, and what it believes will make Watson-like machines better.
But here’s the irony: we, as humans, also look at that unstructured data, and it’s part of what makes fantasy football fun. While the computers turn the unstructured data into algorithms, we put it through our personal system of intuition and decide if it’s relevant or not. We see things not apparent in the numbers that tell us that Teddy Bridgewater might flop in Carolina, or that Josh Allen is ready to break out for the Bills. Once that has gone from the fantasy football game, you might miss it – even if you are winning more.