[M T]

Um, this is how Connelly describes his post-gam win expectancy:

What is postgame win expectancy? It is my look at the key, predictive stats produced by a given game -- a lot of which ends up in the machine that produces my SP+ rankings. In essence, it takes these key stats, tosses them into the air and declares, "With these stats, you could have expected to win this game X% of the time."

Paywalled link: https://www.espn.com/chalk/insider/insid...raska-side

Stanford's post-game win expectancy was 23% against ND and 26% against Oregon State.

For the stats in the game itself (which generally favored Stanford), a Stanford win expectancy of 23% makes no sense.  

So, it must be that he is saying, add the stats from the game into all the other stats he has, and then again predict the outcome of the game (already played).   That would be perhaps interesting early in the season, but not useful at this time of the season (except for maybe changing between 49% and 51%) since the prior games' stats will have the most weight.

On the other hand, if he threw out all stats except for this one game, then his number might be interesting.  But if he did that and came up with 23%, something is broken.

So, I'd say:  "Who cares?  While the pregame expectancy was relevant, the postgame expectancy is meaningless."

[M T]

Um, this is how Connelly describes his post-gam win expectancy:

What is postgame win expectancy? It is my look at the key, predictive stats produced by a given game -- a lot of which ends up in the machine that produces my SP+ rankings. In essence, it takes these key stats, tosses them into the air and declares, "With these stats, you could have expected to win this game X% of the time."

Paywalled link: https://www.espn.com/chalk/insider/insid...raska-side

Stanford's post-game win expectancy was 23% against ND and 26% against Oregon State.

For the stats in the game itself (which generally favored Stanford), a Stanford win expectancy of 23% makes no sense.  

So, it must be that he is saying, add the stats from the game into all the other stats he has, and then again predict the outcome of the game (already played).   That would be perhaps interesting early in the season, but not useful at this time of the season (except for maybe changing between 49% and 51%) since the prior games' stats will have the most weight.

On the other hand, if he threw out all stats except for this one game, then his number might be interesting.  But if he did that and came up with 23%, something is broken.

So, I'd say:  "Who cares?  While the pregame expectancy was relevant, the postgame expectancy is meaningless."