Super Bowl LIII Prediction

Pains me to post this image!
Source: George Martinez

Blog Update

Just a quick update to the blog. I had done a quick update last week that I have been posting daily win projections on Twitter for NCAA Division I Top 25 teams for at least a couple of weeks now. Yesterday I made a break through in automating the entire process so now I have a script that runs automatically every morning at 6 AM CST to scrape the previous days’ game results, update my model, and then apply that model to spit out the probability that the home team will win. So now I have results for every game played on that day as well as the Top 25 teams.

That’s one of the two models I have for Men’s College Basketball games. The other algorithm, one I had stopped using for a few weeks, was also fixed this weekend. I mentioned a while back that, while running 10,000 simulations of the rest of the season, it kept insisting that Loyola was going to finish the regular season as the #1 team in the nation. The problem ended up being, which Loyola!??!? Illinois or Maryland? Yup, once I realized that I didn’t differentiate the two in my algorithm (as well as Miami-OH and Miami-FL), I fixed it and now have something that makes way more sense. Unfortunately that model takes nearly an hour to fit even with a GPU runtime used on Google Colab, so I’ll more than likely only run it Monday mornings to compare against the official Top 25 list.

One more thing, Friday night I scraped the January 2019 Box Office results so will hopefully have a visualization post about that coming soon as well as a breakdown of the movies I saw the first month of the year.

Anyway, let’s get to football, shall we? That’s why you’re here!

Conference Championship Recap

Two weeks ago I had predicted that the Saints and the Chiefs would be the teams that would be playing tonight in the Super Bowl. I really only had one model that had the Los Angeles Rams moving on, but, at only 50.3% of a chance…basically a coin flip. In any case, my Briar scores for all of the models weren’t all that good ranging from 0.251 to 0.384.

Super Bowl LIII Projections

New England Patriots vs. Los Angeles Rams – 5:30 PM CST (CBS)

Obviously neither the Patriots nor the Rams were the teams that my math (overwhelmingly) picked to be in the Super Bowl, but here we are. I’ve found that most predictions for tonight’s game are not all that confident (over 60% probability) for either team (a couple of examples shown below). This basically means, to me anyway, that this is going to be a tight game that could go either way. I imagine that this game will go the same way the Chiefs-Patriots game did two weeks ago: Patriots jumping out to an early lead, then the 2nd half just being a whirlwind of offense with both teams trading points until the very bitter end.

Here is a breakdown of each of my models as well as predictions from various other sources:

Models Proj. Winner Win %
Model 1 (Poissson) Los Angeles Rams 58.85%
ELO Model 2 (HFA=54.8) New England Patriots 52.78%
Model 3 (FO+ELO) New England Patriots 63.33%
Model 4 (Bayesian) Los Angeles Rams 53.43%
Final Projection New England Patriots 50.96%
Vegas New England Patriots 56.15%
Cynthia Frelund Picks Los Angeles Rams 53.2%
Football Outsiders New England Patriots 50.1%
FiveThirtyEight New England Patriots 53%

The Top 5 models above are my own, and the bottom four are Vegas Money Line Implied Probability Odds from Pinnacle, Cynthia Frelund’s pick from the NFL, Football Outsiders’ pick, as well as FiveThirtyEight’s pick. As you can see all but one model have projections less than 60%. Only my own personal model that takes into account a tweak of the ELO model from FiveThirtyEight as well as a handful of statistics from Football Outsiders is “confident” of a Patriots win.

As far as points scored, on Cynthia Frelund’s podcast, she projected the Rams to win 29-28. My personal projections have the Patriots winning 30-29. Either way we both have the Rams scoring 29 points which I thought was interesting.

I’m hoping that the game is rather entertaining to watch for sure although I may not watch the entire game live. When it comes to Super Bowls, I tend to record it, then later fast forward through the game to watch the commercials. It’s a thing. At my home today we’re doing an old fashioned BBQ for the game.