[CompSci87]

[quote author=Griffin78 link=topic=8426.msg68728#msg68728 date=1379867337]
I agree with martyup Guessing the winning margin isn't as good a measure as guessing the final score(s).

Should we calculate the difference between the predicted and actual Stanford score and add the difference between the predicted and actual opponents score?

I agree as well.
When I have time, I'll put up new rankings using |PredStanford-RealStanford| + |PredRival-RealRival| to order the list instead, and people can decide which metric they like best.
[/quote]

I think this formula is the right idea, and the penalty for wrong winner is good too. When I think of predicting a game score, I think of predicting who will win and the number of points each team will score -- not predicting the spread and throwing in the over/under as a tiebreaker.

I gather that in addition to the above, you're using the spread as a tiebreaker, right? That seems good too.  For example, if the true final is 13-10, a prediction of 16-13 feels closer to correct than 16-7.

[CompSci87]

[quote author=Griffin78 link=topic=8426.msg68728#msg68728 date=1379867337]
I agree with martyup Guessing the winning margin isn't as good a measure as guessing the final score(s).

Should we calculate the difference between the predicted and actual Stanford score and add the difference between the predicted and actual opponents score?

I agree as well.
When I have time, I'll put up new rankings using |PredStanford-RealStanford| + |PredRival-RealRival| to order the list instead, and people can decide which metric they like best.
[/quote]

I think this formula is the right idea, and the penalty for wrong winner is good too. When I think of predicting a game score, I think of predicting who will win and the number of points each team will score -- not predicting the spread and throwing in the over/under as a tiebreaker.

I gather that in addition to the above, you're using the spread as a tiebreaker, right? That seems good too.  For example, if the true final is 13-10, a prediction of 16-13 feels closer to correct than 16-7.