NFL Machine Learning Picks: Super Bowl 57
Kansas City +1.5
The biggest variable is Mahomes high ankle sprain and if he’ll be able to survive the whole game without limping. Playing against the outlier #1 team in sacks who hurt two QBs in the last game, the likelihood of him getting injured increases significantly.
Assuming he can play and run around for the whole game, even if it’s at 75%, here’s my analysis. Otherwise, if he gets injured, it’s over just like the 49ers.
Offense and Defense
Both teams scored an average of 27.8 points per game and are ranked 1st (KC) and 3rd (PHI) in the league. The Eagles did it mostly on the ground and the Chiefs with the pass. The Chiefs lead the league in passing DVOA while the Eagles lead the league in running.
The biggest matchup is KC’s #1 passing offense against the #1 passing defense. The Chiefs could decide to build a bigger impacting ground game for this game as the Eagles rushing defense is below average. Feed Pacheco and McKinnon and this #9 rushing offense could find success.
The Eagles on the other hand will be able to move the chains on the ground. Only with good preparation will KC be able to stop the run game. Now a lot of the Eagles run game comes from Hurts and he will be facing the 2nd best pass rush in the league. The Eagles have great wide receivers that can exploit the Chiefs secondary if Hurts has enough time to throw the ball.
The advantage of the passing game is a security blanket if the team is down. In that case, there’s an advantage for KC. On the other hand, if KC can’t find receivers or are getting rushed, then we’re back in 2019 when the Bucs stopped Mahomes by running through the offensive line. This year, the Chiefs were 5th in the league in allowing the least sacks while the Eagles were 21st. That could be trouble for the Eagles and a potential relief of stopping the mad rush of the Eagles.
Strength of Schedule
It is important to note that the Eagles had one of the easiest schedules in the league. It was predicted at the beginning of the season and after the season they remained with one of the lowest strength of schedule in the league. The Chiefs were predicted to have the 2nd toughest schedule and finished 8th. They have also had a much tougher run for the playoffs after beating the Jaguars and the Bengals, than the Giants and a team with a 5th string QB.
The quarterback being the most important position, I’m a fan of backing the team with the better QB. But who’s better, Mahomes or Hurts? Ok, that shouldn’t be a question. Mahomes numbers for his first 5 five years in the league are breaking all kinds of records. Mahomes needs this ring so the conversation about him being the potential new goat (long way to go though) solidifies.
Hurts time will come, especially with that receiving squad, but they might need another shot at it. His passing yards aren’t impressive. He threw for 3700 yards, 22 TD and 6 INT, rarely throwing over 300 and a few times under 200. Experience counts and there’s a chance he makes too many mistakes like last year when they lost against the Bucs, they were down 31-0 at some point and he threw 2 INTs.
Both great coaches with different styles. Syrianni is more aggressive and risk taking while Reid game plans to keep the score close and end it when needed. This is Syrianni’s 3rd game in the playoffs as a head coach, he’s 2-1. Reid is 20-16 in the playoffs and 10 of those wins with the Chiefs. His trademark is preparation and has a ridiculous 28-4 record after a bye week.
Machine Learning Models
I understand that it is a bit of black box even for me. I know the tools being used and ingredients to make the outputs, but why precisely do my models like the Chiefs over the Eagles is hard to tell. Could be that the differences in some offense and defense metrics benefit the Chiefs more than the Eagles. It could also be because they beat harder teams. Could also be that the differences in point (line), after an outlier effect minimization, also benefits the Chiefs.
Whatever the reason, all my models point at the Chiefs and some have them winning by 10! On a weighted average, as the output below shows, I predict they win anywhere between 3 and 7 points.