Mid-Week Metrics Honors the Fallen Coach

Any way you can give us your predictions on the outcome of the LSU v. Georgia game and the MSU v. Wisky game?

And if you're feeling ambitions and your models permit it, can you predict M's chances of getting into the Sugar Bowl?

I find it hilarious that THE KNOWLEDGE and The Mathlete have back-to-back diaries.  Two polar opposite techniques of predicting the future.  Love them both.

I agree with the sentiment that the RB position is overrated.  Especially these days with the rules so much in favor of passing attacks, the quarterback position is the clear king.  The scheme and strategy/game plan is also important, which is why I believe that coaches are very underrated.

Coaches always talk about RBs "getting stronger as the game goes on" when you have one back getting the bulk of the carries.  Is there anything at all to this?  Is it just the defense tiring (so would any other back that comes in have equally improved stats)?

Is there anything to having a quarterback in a rhythm/taking him out of rhythm?

Are 4th down plays in critical game situations more or less likely to be successfully converted than equivalent down and distances in non-critical situations?

Mathlete, you've done a great job all year and I really, really appreciate the content you add to the blog.  Your predictions throughout the course of the season has always calmed my nerves a bit on Saturday afternoons.

Denard had the 5th best game of any QB this year at +24.

Do you calculate this data for every player in the country? How do you do it?

Thanks for all you do - this is by far my favorite diary.

Any chance when you plot progressive win odds that you could add confidence intervals?  

Any chance when you plot progressive win odds that you could add confidence intervals?  

I see what you did there

Sir Mathlete, 

Do you think an adjustment needs to be made to your formula's to better weight the DEs/LBs? They seem to get HUGE numbers based on what's probably lots of tackles and/or sacks? I find it a little head scratching that a good to great DE is more valuable to winning a game over a good to great quarterback (see: Russell Wilson). 

Thanks I'll hang up and listen. 

This wasn't deserving of it's own thread...but does ANYONE know what's going on with the UFRs?! I've been checking daily looking for Brian's work and haven't heard/seen anything.

Did you use only statistics from conference games to determine your all Big 10 team?

As a systems student I am curious how some rather fuzzy information (case: Brian Kelley's red face) impacts your prediction, if at all. I am pretty sure it is a big deal but I have no idea how that can be included.

Good read, as always.

Mathlete,

How comfortable are you with your team rakings based on the system that you have developed? and would there ever be a push from you to try and be added into the BCS mess?

Also, after this weekends games, could you post what your top 25 teams are?

Great stuff

Couple questions

Who would you vote for Big 10 Offensive Player of the year....despite the numbers above, I would have to cast my vote for Montae Ball, with Denard second. I dont, but from watching all year, I think Russel Wilson is a little overrated and he wouldnt crack my top-2

Also, will KState cover that big number (-11, i think) against Iowa State....the pickings are slim tomorrow with so few games and I need a winner. So, give me at least one!!

Thanks!

That progressive win odds chart probably looks like an EKG of M fans during the Ohio game.  For such a close game with all the pressure, there were a lot of up and down moments.

Dear Mathlete,

Thanks for a great entry. I really enjoy reading them.

Two thoughts and I apologize if you had addressed them earlier in a previous post.

1. Plays not made are also important but are much harder to quantify. In the OSU game everyone saw that Posey was wide open at the end and had Braxton Miller not overthrown him he probably would have scored leaving the Wolverines with about 1:30 to win the game. Assuming they would start from the 20 they would have to get down at least under the OSU 30 to have a confident field goal attempt. Certainly not impossible and after the ND game UTL we know that miracles do happen. But that single play would have change the possible outcome from a high UM probability to a pretty good OSU probability. Those aren't factored in. Plays not made are always going to be tough. In the 1997-1998 Rose Bowl Charles Woodson was within a hair's breadth of sacking Ryan Leaf several times but it didn't happen. Had that happened the game wouldn't have been as close. Is it possible for you run a simulation of what the odds would be if Posey had scored? If you are familiar with computer chess programs look at Fritz (by Chessbase) it provides a similar analysis of who is "ahead" in the game and one can see graphically in much the same way you depict where the critical turning point occured. It has one huge advantage in that it can analyze the alternative lines of play.

2. Which leads to the second point and suggestion. Others have also noted that time and situation affect the effect of any individual play. Any single play doesn't exist in isolation. Failing to gain yardage on 2nd down leading to 3rd and long is a common example. What I'm curious is about is whether your analysis can shed light on the relative value of points and leads during the game. In the Nebraska and OSU games the visitors scored first but they did so very early on the game. So one could surmise that a +7 score for a team with  50 some minutes of time isn't crucial. Being +1 with 0:30 might be conclusive unless the ball is within easy field goal range for your opponent. The thought occurred that it might be another way of looking at how well a defense plays. A great defensive team could hold a small +lead over a longer period of time than a poor one. In traditional football terms, the great defense needs only to be ahead to win most of the time and not two scores or three scores. Is that calculable? I suspect this season our defense would do well. We took the lead in the 3rd quarter and while OSU did score another 10 points, we never trailed.

Again a great diary and hope that you'll keep it going!

Great stuff.  I'm still surprised a bit about Wilson having such great numbers, since he seems like the weak link of that Wiscy attack (maybe him along with the inconsistent WR play). 

For the OSU win odds chart, can you please plot a couple extra data points at the end of the game?

I know these go against the "result of the play" method but I'm dying to see what would show. 

Ex:

Part 1: 

1st and Goal at OSU 4             Denard Robinson rush for a loss of 1 yard to the OhSt 5.             

2nd and Goal at OSU 5            Timeout OHIO STATE, clock 2:41.                 

2nd and Goal at OSU 5            Fitzgerald Toussaint rush for 4 yards to the OhSt 1.*             

3rd and Goal at OSU 1            MICHIGAN penalty 10 yard Holding accepted.                      

3rd and Goal at OSU 11          MICHIGAN penalty 15 yard Personal Foul accepted.              

*Add a play here for “Fitz TD”.  Then “undo” it to the actual result of the replay overturn call.

Part 2:

2nd and 10 at OSU 20            Braxton Miller rush for 4 yards to the OhSt 24.             

3rd and 6 at OSU 24               Timeout OHIO STATE, clock 1:48.                 

3rd and 6 at OSU 24                Braxton Miller pass incomplete to DeVier Posey.*             

4th and 6 at OSU 24                Braxton Miller rush for 7 yards to the OhSt 31 for a 1ST down.             

*Add a play here (and this is fudging a little bit more) for OMG Posey is behind the defense and a 60yd bomb is in the air and he should score on this play.  Then “undo” it to the actual result of incomplete, wow we’re lucky that Miller missed that throw.

Just sayin'...

Hey Mathlete, in the 11/23 Wall Street Journal, Accuscore said Michigan had a 60% chance of winning, based on 10,000 game simulations.  Your initial probability had Michigan around 71-72%.  What methodologies do you and they use to come up with these initial probabilities?  Have you tracked yours versus theirs to see who is more accurate?