splat_edc

​

Teams are sorted by Pro-Football-Reference's "SRS"/"Simple Rating System" converted into winning percentage using the formula laid out here. Pythagorean winning percentage is based on points scored and points allowed per game.

Week 11 Predictions:

All the various official and unofficial power rankings have inspired me to start tracking my own rankings and predicting matchups each week. The column on the far right is based on a weighted average of each teams' actual win percentage, their pythagorean win percentage, and everyone's favorite concept, "regression towards the mean" plus an adjustment for home field advantage. Since I started keeping track in week 6, I have correctly picked the winner 63% of the time and Vegas is at 68%, which is pretty close. Unfortunately, in head-to-head disagreements with Vegas I'm 5-9 which is pretty rough.

As with most weeks, power rankings and vegas agree basically across the board. Only differences right now are MIN@DEN and PHI@KAN. My numbers and the power rankings are backing Dobbs and the Vikings, while Vegas goes with the home team Broncos. Maybe the Broncos have started to right the ship, but that MNF win was a weird one and Dobbs has been impressive with Minnesota thus far. Current line is -2.5 Denver, which basically means 50/50 plus home field. To me that feels appropriate, though the power rankings would seem to imply there's a wider gap between the two teams.

The other disagreement is Chiefs and Eagles (spread is -3 for KC). Another coin toss albeit for very different reasons. My numbers agree with the power rankings that Philly is the slightly better team, but then home field advantage tips the scales in favor of Kansas City aligning me with Vegas. Either way, should obviously be a good one.

My numbers also have Green Bay as favored vs the Chargers, at odds with both Vegas line and the power rankings. Not feeling good about that one.

First off, thanks for putting this together. Interesting stuff to dig into.

For the two-tailed z-scores does that mean looking for scores greater than/less than +/- 1.96 or am I thinking of something else?

On the success rate numbers, are those only for pass attempts or do they include things like QB scrambles? I noticed that Mahomes has a decent success rate but is below average in these passing metrics. Could that be due to his ability to pick up yards/first downs with his legs? I’m assuming success rate doesn’t distinguish between air yards and YAC, so for somebody like Mahomes (Ridder has the opposite thing going with bad success but decent CAY) can we attribute the success to YAC or is that an unjustified leap?

It’s interesting to see where the pure win percentage approach diverges from the margin of victory approach that drives football-reference’s SRS. Top 5:

(1) BAL

(2) SFO

(3) DAL

(4) JAX

(5) BUF

The Eagles are down at 9 in SRS. The biggest differences are the Texans going from 8th in SRS to 19th in the colley matrix and Pittsburgh which jumps from 20th SRS to 10th. San Francisco jumps 9 spots up from 11th in colley to 2nd in SRS. Jets also move nine spots in the opposite direction (22nd SRS and 13th adjusted win%).

The correlation of the ordinal ranks is .81, so there’s still broad agreement between the two rankings but the disagreements are interesting.