Elo is weird for games against non-FBS opponents in cfbd_game_info()

[Neil-Paine-1]

I've noticed that an FBS team's pre/post game Elo ratings in cfbd_game_info() are wonky for games against non-FBS opponents. The two values are identical, which makes sense (as ratings don't change after a game vs. an unrated team), but those values are not the final Elo from the previous game against an FBS opponent. And then, the pregame rating for the next game against an FBS opponent reverts back to what the postgame rating was from the last game against an FBS team.

This might seem like no problem, except that the Elo ratings called from cfbd_ratings_elo() use the most recent postgame ratings, which are broken for teams that just played non-FBS teams. Lemme give an example:

Penn State was a 2087 Elo after beating FIU on Sept. 6. Then they played non-FBS Villanova -- and their pre/post-game ratings are listed as 2152 for some reason. The next game is against Oregon, and their pregame rating reverts back to 2087, where it was before the Villanova game. This seems to be happening whenever an FBS team plays a non-FBS team.

Thanks!

[akeaswaran]

Hi Neil -- sorry for the delay here. Took a look at this just now, and it seems like this is how the data is coming in from collegefootballdata.com: I can see this trend week over week in the raw data on the website (Analytics > Data Exporter > Games > Retrieves historical game data > 2025 / Penn State). You may want to flag this with Bill there on BlueSky (@collegefootballdata.com) or email (admin@collegefootballdata.com).