When I was growing up there wasn’t a day that would go by without some sort of activity regarding sports in my household. Whether it was playing countless sports outside with the kids in the neighborhood or me imagining myself as a world champion in my parents’ living room, sports always seemed to be on my mind. Those dreams eventually allowed me to play baseball in college, but as with most athletes, there comes a day when your playing days are over and the real world shows up at your door.
When I was no longer playing any sport, my love for them started to shift to the statistical side of things. But I guess you could say that my love for the statistical aspect of sports started way back in 1999 in a Lemont Elementary School kindergarten classroom in State College, Pa., when my teacher allowed me to tally which MLB teams won the previous night. Although that’s one of the simplest forms of sports statistics, my post-college I took it upon myself to further my knowledge.
I set out to learn as much as I could, with the goal of being able to know who were the best teams in each sport and then see how my projections and assessments would fare in the sports betting markets. These projections started out at the team level for all the major sports and eventually grew into player-level projections for sports like NBA and MLB. Through years and years of tweaks and improvements, the use of Microsoft Excel to learn Python, the models I currently have look very different than when I started. But it’s been all for the better.
If you’ve read any of my content before, you know that most of my models are used for sports betting purposes. But if you’re not into sports betting, that’s totally fine; they’re still of use. How many times have you been in conversation with friends about who the best team or player is in a sport? I know I’ve done that more than one should. I wanted to expand my models to do a better job of telling this story, which led me to build season simulators for each sport so I have outputs for any team to make the playoffs, win their division, and of course, to become a champion . Based on my model outputs, I can provide my model’s projection for a team to win the Super Bowl by simulating every game.
For transparency purposes, let me go deeper into my process. I try to use the most granular data to get a more complete picture of a certain sport. This means, for example, play-by-play data for NFL and college football or plate appearance data for MLB rather than individual game data. This allows me to strip out any garbage time plays that aren’t predictive. From there, I’m able to roll up any metrics that I find to be useful and create an overall team projection, which I then put into the simulator.
Although the process of forging a team projection mostly remains the same across sports, I am constantly looking at how my model is performing against the sports betting markets. If I start to find an inefficiency or if any new data is available that is deemed to be valuable, I’ll adapt and make changes if I see fit. So just like everyone’s favorite coach or player, I too am making adjustments to become better at any moment. While it takes a lot less physical activity for me to do that than it does a team, my goal is to make my model as accurate as possible, and that takes constant supervision.
A lot of data-focused individuals get a lot of flak in the sports content world because “the game isn’t played on a spreadsheet”. Some of that may be warranted, but I like to think data-driven content is the same as traditional content but through a different lens. I keep an open mind to different practices and am always willing to adapt. And although my models will be ~95% of my beliefs, I know there will be shortcomings, and outliers do exist. But this doesn’t mean I always hate your team, it’s just that the data does. So yes, the game isn’t played on a spreadsheet, but the data certainly helps tell the stories that we all want to hear in the world of sports, just in a different way.
As someone who loves all things sports, I never get tired of debating, watching and just experiencing the ups and downs that different sports all have to experience. My goal is to use my models to tell a story about why this is happening or why this isn’t happening to better experience sports (and hopefully make some money along the way). Although my five-year-old self’s dream of becoming a world champion are all but over, I think I lucked out and couldn’t be happier to progress from tallying wins and losses for the 1999 MLB season in kindergarten to creating content for all things sports at The Athletic.
(Photo: Kirby Lee / USA Today)
.