Back in late September last year, when the Cowboys were 1-2, I took a look at the win probabilities for all remaining Cowboys games as calculated by the ELO ratings published on fivethirtyeight.com.
The results were surprising, as the ELO win probabilities at the time predicted a 10-win season for the Cowboys. This prediction did not sit well with parts of the fan base at the time, and were summarily ridiculed.
- “ELO....extremely long odds?”
- “You sure that shouldn’t be Elol ratings?”
- “If this algorithm was piss tested, its urine would have stems and seeds in it.”
- “Did Tony Romo come out of retirement and sign a one-year deal with the Cowboys?”
- “Not sold on ELO picking football games, but I’m a fan of their music.”
At the time, the rampant pessimism in Cowboys Nation did not allow for a universe in which the 2018 Cowboys would have a winning record, and even proponents of a multiverse would have been hard-pressed to find a bubble universe, a daughter universe, a mathematical universe, or a parallel universe in which the 2018 Cowboys were en route to a winning record.
Yet here we are, the Cowboys won the division with a 10-6 record (albeit with some unforeseen Amari Cooper-magic), and even won a playoff game. So let’s go back to that ELO well for the 2019 season, but first, some background:
Elo ratings estimate each team’s skill level using only the final scores of previous games and the locations of each game. That skill level is translated into a points system where an average team has 1,500 points, and teams usually fluctuate somewhere between 1,300 and 1,700 points. The Cowboys for example are currently ranked ninth overall with 1,547 Elo points.
Using those points, the Elo system can calculate an expected point spread and win probability for the next opponent.
The Cowboys’ first opponent this season, the Giants, are ranked 28th with 1,424 points, which gives the Cowboys an Elo point spread of -7.5 and a win probability of 74%. And what can be done for the Giants can be done for all of the remaining 2019 opponents. Here’s what that looks like, according to fivethirtyeight.com:
The title of this post gave it away early, the Elo ratings system currently sees the Cowboys finishing the season with 12 wins. How’s that possible you might ask?
- They win every single home game: The Elo system likes the Cowboys at home, and has them winning all eight contests, including the two games against higher-rated opponents (Rams & Eagles).
- They split their road games: The Cowboys are projected to lose on the road against the four higher-rated opponents they face: the Saints, Patriots, Bears, and Eagles.
Of course, what I did above is simply add up the individual game predictions to arrive at the 12-4 prediction, and there are a lot of awfully close games included that could swing either way. fivethirtyeight.com also offers a full-season projection that doesn’t simply add up wins and losses, but also considers probabilities. And that look results in a slightly less flattering 9-7 record projection.
Is 12-4 too high? Perhaps. Is 9-7 too low? Perhaps.
Only one way to find out, and it starts on Sunday against the Giants.