The NFL season is right around the corner, which means that the few weeks prior to the start signifies a time of hardcore analytics for the competitive fantasy football player like myself. I’ve been a frequent and competitive fantasy football player for about 15 years now and I’ve come to realize that the logic that most of us use to select our draft picks really mirrors how an organization should leverage data for business decisions.
In business today we hear a lot about big data and how organizations are embracing the notion of data-driven decisions. Since there is so much data available, most people think that everything and anything can be calculated. The extent of data’s impactfulness today has built demand for the power of data science, where there are advanced statistical modeling that lead to predictions. Because of this, the perception of data-driven decisioning allows one to make the best decision possible. With data-driven predictions, you can make intelligent decisions on a variety of things, whether it’s personalized targeting of your advertising campaigns or optimizing an e-commerce site to increase conversions for example. Just think of the scene from the Minority Report where Tom Cruise runs through a shopping area and all the stores are able to identify him and then accurately personalize their marketing content and ads. Predictions can also be seen all over the sports world. Vegas sportsbooks are constantly using big data to predict scores (for gambling purposes). When you look at how inaccurate they really are, it shows that a singular reliance on numbers doesn’t necessarily get you the truth.
Nate Silver a respected statistician recieve his claim to fame by using analytics to correctly predict the 2008 presidential race. Since then, he has moved on to bigger things, such as being the chief editor for ESPN’s FiveThirtyEight.com site. There’s a general public assumption that when he makes a prediction, it will most likely be right. When you dive into his big predictions in the sports world, you will realize that he really hasn’t been on point. Super Bowl XLIX matched the New England Patriots against the Seattle Seahawks.
As a Boston sports guy, I was obviously rooting for my Patriots, and I soaked in all the media surrounding the game. One thing that stuck to me was Nate’s prediction that Seattle would win. I’m happy to say that he failed on that prediction. It should also be noted that Silver picked Carolina to win in the following Super Bowl. Wrong again. Perhaps the ultimate event to test your big data decisioning effectiveness is the NCAA tournament. Again, no matter how much data you have on the teams’ performance, history, players, etc. today, you still can’t produce a perfect or even near-perfect bracket. It should be noted that FiveThirtyEight.com picked Villanova to win it all, which was not the case.
So, when this is applied to fantasy football, big data is utilized to predict a given player’s performance through the course of the season. Fantasy sites have done a great job at finding and utilizing so many data points to algorithmically produce a ranking system. However, no matter how much data you have, the rankings never mirror real-life standings by season’s end. It’s fair to say that your first round pick is critical to the success of your fantasy team, and that is pick you can’t screw up. If you look at the number one ranked players in the last five years below, none of them have finished in first place as they were predicted to do.
One criteria that is not sufficiently factored into the ranking is the probability of a player’s susceptibility to injury. You can argue that this is due to pure randomness that can occur at any moment in a concussion-inducing sport. Therefore, it is the same probability to everyone. However, we know in reality there are just some players who are more durable. We also know that some work harder, some play smarter, some have a bad history, and some just have a knack for avoiding serious injuries. There should be a formula for factoring these things into the predictions we are making.
When it’s time for me to make my draft pick, I can’t solely rely on the general ranking system of players. I most definitely use those statistics as part of my decision-making, but I am not 100 percent reliant on it. Since the statistical model is “incomplete” in my eyes, there’s an element of the decision that’s going to come from the gut. Hardcore analytics practitioners are probably shocked that an analytics professional would say this, but that’s the reality.
At the end of the day, the decision is based on what I know and what I feel, mixed in with the data at hand. This mentality should be mirrored in business decision-making. Unless the data is clear cut, you have to make the appropriate decision given all the circumstances. In other words, one’s draft pick selection needs to mix statistical rank with a personal assessment toward a player’s injury propensity. This is the essence of making data-informed decisions as opposed to the misleading nature of data-driven decisions. The data is not always the be-all and end-all.
Former NFL quarterback Steve Young once said that although the numbers don’t lie, they also don’t always tell the truth. A human touch is still needed to fully interpret the data’s story. You may have a scenario where a site visitor has incurred 20 page views in a particular site section. But were there so many page views because the visitor was engaged or just downright frustrated? A data-driven approach would steer additional visitors to that area due to the data. A data-informed approach would consider other information, some of which may not be quantifiable…such as experiencing the site yourself, focus groups, or even surveys. With more information, you can build your story and determine the appropriate solution.
Ending back on fantasy football, if I had the first pick in my fantasy draft this year and I look at all the great players with the top pick, I think I’m still taking Antonio Brown. Sigh.