Fantasy Football is Teaching Data Fluency

 

Fantasy football season is here again (along with the actual NFL season). I thought it a good time to share a section from our upcoming book Data Fluency, scheduled to be published in October through Wiley and with Nathan Yau of FlowingData as editor. In this excerpt, we suggest that Fantasy Football has taught an enormous audience to understand the language of data:

It may not be a stretch to say more Americans have learned about data and statistics through fantasy football than every college statistics course in the country. Each week, some 19 million NFL football fans spend their Sundays meticulously setting team line-ups based on statistical projections, historical patterns, and analysis of week-to-week variance. The couch potatoes who once relished on-field hits and in-game strategies now spend an average of more than eight hours a week diving into the data of the sport.

For the uninitiated, fantasy sports let fans play the role of team owners and managers by picking players for their own fantasy team and making weekly roster decisions. As the action plays out each week on the field, fantasy owners collect points against other competitors within their fantasy leagues. To win, fantasy owners quickly realize that success often depends on studying player and team performance data closely.

Here are a few ways that NFL fantasy players incorporate data into their thinking:

Variation in Player Performance

The best fantasy owners understand the nature of week-to-week variance and its relationship to earning points. For example, touchdowns generally earn a fantasy owner six points; but touchdowns occur rarely and can fluctuate wildly. In contrast, the number of touches players receive may be a better indicator of how much the team is using them and their opportunity to provide the owner with points. Because consistent performance matters, successful owners often focus on players with more stable predictors of success (for example, touches) versus more sporadic events (for example, touchdowns).

Rankings Can Be Misleading

Fantasy football cheat-sheets offer rankings of players in every position. These ranking mask the differences and dispersion of expected performance. For instance, the top running back may be expected to perform 20 percent better than the second rated running back, who in turn is only expected to score 5 percent more points than the third through sixth rated running back. The data shows that players often cluster into tiers of performance. This statistical understanding was publicly explained by Boris Chen who stated that “players within a tier are largely equals. The amount of noise between the ranks within a tier and actual results is high enough that it is basically a dice roll in most situations.” This concept has been widely adopted by fantasy owners as a player drafting strategy. 

The Only Constant Is Change

The worst fantasy football owners are stuck in the past and pick players and teams that they have relied on in the past to generate points. That is, they fail to update their assumptions about the best teams, players, and trends. Following the data closely reveals when certain players have gone past their prime and when teams that once had high-scoring offenses can no longer put up big points. Clinging to past success may be a formula for disaster because the only constant in fantasy football is change.

Context Fills Out the Picture

Data viewed in isolation can be deceiving. Say, for example, that your top wide receiver scored only one-half the number of points that he scored on average in a season. Is this a new and troubling trend? Should you trade? A little research might reveal that he matched up against one of the league’s top cornerbacks, or his quarterback was knocked out of the game, or perhaps he tends to perform poorly in cold weather, away games. These environmental factors make a difference with respect to outcomes. Performance data cannot be understood in isolation—context matters.

So how did fantasy football create legions of fans who have developed a specialized dialect of data fluency? It has been a combination of education, effective data presentation, common data conventions, and incentives. Fantasy football owners have been taught how to use data to their advantage through the efforts of the NFL, ESPN, Yahoo!, and a cloud of other websites dedicated to football analyses. Organizations like Football Outsiders built new media businesses around data modeling and projections of player performance. 

Leading online fantasy football sites like ESPN and Yahoo! have been aggressive in pushing data and data visualizations to their users. These sites include trend charts for every player, drive charts, player comparison graphics, and predictive models for estimating game outcomes.

The educated fantasy football community is also highly engaged with the sport. The community loves football! The fantasy league has provided a whole new (and rewarding) dimension to its fandom. No longer is it tied down to rooting for a single team—instead, the whole league becomes fodder for its attention as it picks and chooses players from each of the 32 NFL teams. In addition, the fantasy football industry has coalesced around consistent formats for leagues, points, and key metrics. Terms like PPR, running back by committee, waiver wire, and flex are well understood, facilitating conversations among league owners. And with $1.18 billion bet in fantasy football leagues annually and a passionate fan base, fantasy owners have huge incentives to make informed decisions. When money or bragging rights are on the line, individuals invest time and energy into developing the skills and abilities to become data fluent.

In short, these factors have brought data fluency to the masses. Millions of fans have learned how to read charts, grasp basic data concepts, and allow deeply embedded data to inform how they make decisions—all critical skills associated with quadrant one in our framework.