This hits hard! My NFL model (trained on data since 2018) says that the effect of home field advantage is weak and not statistically significant (after accounting for team skill). It has done very well in the regular season but had some heartbreaking bad beats in the playoffs, which are represented in your table. Maybe another season or two and I'll consider adding a playoff HFA variable
I love this analysis. What are the sample sizes by sport? I imagine it's largest for NBA and NHL and still very small for the NFL
Good question! Here are the total playoff games in the "post-Covid" era (as of Monday AM):
NHL: 296
NBA: 283
MLB: 124
NFL: 35
So yeah, spot-on in your guesses.
This hits hard! My NFL model (trained on data since 2018) says that the effect of home field advantage is weak and not statistically significant (after accounting for team skill). It has done very well in the regular season but had some heartbreaking bad beats in the playoffs, which are represented in your table. Maybe another season or two and I'll consider adding a playoff HFA variable