Did an analysis of NFL home field advantage for 2020. Bottom line is that the 4 methods I used to solve for it resulted in values of 0.03, 1.18, 0.56, and 0.11 points per game.
The first 2 values were by using pre-game FPI team strength estimates (which accounts for QB changes) to estimate actual game score difference. 0.03 using L1 optimization (abs error), and 1.18 using L2 (squared error).
The other 2 values were by backfitting a team strength point value to estimate game score differences. (Similar to SRS and I think one of Sagarin's models.) 0.56 using L1 and .11 using L2.
One thing to be aware of is that a short 256-game schedule can result in home and vstr imbalances. Some years the home teams are better on net, some years the vstrs are better. This year was balanced--home tms were better by .09 pts per game absent any HFA, according to FPI.
Not to mention usual year-to-year random variance.