It was brought up in another thread by ChickenMilk that there were rule changes to NCAAF games made prior to this season starting. A description of the changes can be found here:

http://deadspin.com/5041480/explaining-the-new-college-football-clock-rules

It appears from the article that the end result of the changes is net loss of 4 to 5 plays per game. I've noticed that the vast majority of the predictions this year are for the over and that the O/U results are way down from last year. So this begs the question, are the SIMS accounting for the new rule changes?

Post edited by: TPiranha, at: 2008/10/12 03:37

Taken from the NCAAF DLR:

Last year there were 336 over predictions and 319 under predictions or 51% for overs.

This year there have been 194 over predictions and 114 under predictions or 63% for overs.

The record on under predictions this year is 56-54 or 51%.
The record on over predictions this year is 86-105 or 45%.

The record on under predictions last year was 179-140 or 56%.
The record on over predictions last year was 189-147 or 56%.


Clearly something is wrong here. This is beyond normal statistical variance.

It's hard to say what the change comes from. Two years ago the totals were a problem. Last year they rebounded, only to fall back this year. Is this a result in the new rules, or is it a result of Vegas altering their lines? Because we can't be sure, we're not going to adjust the simulations based on a guess as to the reason why the totals are down this year.

"we're not going to adjust the simulations based on a guess as to the reason why the totals are down this year."

You wouldn't be adjusting the SIM based on a guess. You'd be adjusting the SIMS because a rule change was made to the clock rules of the game. Time is an absolutely vital determinant of the games score.

To put this another way, pretend hockey decided to play two 1/2 periods instead of 3 or baseball decided to play 8 innings instead of 9. Are you saying the SIMS wouldn't be changed to reflect this?

"To put this another way, pretend hockey decided to play two 1/2 periods instead of 3 or baseball decided to play 8 innings instead of 9. Are you saying the SIMS wouldn't be changed to reflect this?"

That's not the case here. Football is still the same length. Sure, the clock runs, and that results in 4-5 fewer plays. But is there proof that this translates in to a significant decrease in points? You're focusing on the clock rules being adjusted, but how do we know the teams aren't also adjusting to these rules? Maybe they take less time than normal to call a play, or to drive down the field.

There is no statistical evidence on the impact the clock rules make on the length of the game. There is also no statistical evidence of how the coaches have adjusted to the new clock rules. Therefore, any changes to the simulations would be based purely on assumption as to how much the new rules change the final score.

We provide the simulations with an effort to give an unbiased prediction on the game. From there you are encouraged to provide your own analysis. Say for example there's a game with an O/U of 48 points, and the sims have the game at 49 points. If you feel the new clock rules aren't accurately represented in the simulations, then you might look at the 49 point forecast and determine that with your analysis of the new clock rules, the total score would be less, which would change the AccuScore prediction.

I'm sure if you ask 10 people, you're going to get 10 different answers as to how the new clock rules change the outcome of the game. Because there is no concrete answer, there is no way to make adjustments, unless we used our own assumptions, which would go against the spirit of the simulations.

So are you telling me that Accuscore's simulations are using different clock rules than the clock rules being used in the actual games that are being played?