[PrinceLightfoot]

Wow, no need to ask this year whether there is interest on doing the game again!

Hi again all.
I just got back from a few months in Europe (where, yes, they also have internet, but I was more busy), and I'd be happy to compile the results again. Though I see that PrinceLightfoot has already put together a beautiful table  ;)

My c++ code is based on a class that takes a text file similar to that table, with the results of each week, and implements the scoring system described here. Basically, |PredStanford-RealStanford| + |PredRival-RealRival| with a few details.

The one thing that I wouldn't do again were the comments as tiebreakers. To make the table readable, I first had to summarize them, and then manually tell the code which comment was closer to reality (it'd be a great PhD thesis to teach c++ to read a few articles summarizing the game, and pick the best comments based on that).

Apart from that, there were some gripes with the scoring system (we started with using only the spread, and changed midway to use those absolute differences), but I think is more or less ok.

If anybody prefers to take of this, I'm happy as well.
My prediction for this week is 35-9, an unsatisfying win where the defense is decent but lots of missed opportunities in offense. More yards will be earned through the air, than rushing.

Welcome back, Gauss! That chart is all you - I just changed inputs :)

[PrinceLightfoot]

Wow, no need to ask this year whether there is interest on doing the game again!

Hi again all.
I just got back from a few months in Europe (where, yes, they also have internet, but I was more busy), and I'd be happy to compile the results again. Though I see that PrinceLightfoot has already put together a beautiful table  ;)

My c++ code is based on a class that takes a text file similar to that table, with the results of each week, and implements the scoring system described here. Basically, |PredStanford-RealStanford| + |PredRival-RealRival| with a few details.

The one thing that I wouldn't do again were the comments as tiebreakers. To make the table readable, I first had to summarize them, and then manually tell the code which comment was closer to reality (it'd be a great PhD thesis to teach c++ to read a few articles summarizing the game, and pick the best comments based on that).

Apart from that, there were some gripes with the scoring system (we started with using only the spread, and changed midway to use those absolute differences), but I think is more or less ok.

If anybody prefers to take of this, I'm happy as well.
My prediction for this week is 35-9, an unsatisfying win where the defense is decent but lots of missed opportunities in offense. More yards will be earned through the air, than rushing.

Welcome back, Gauss! That chart is all you - I just changed inputs :)