[CardinalSagehen]

6.  We went one of four on fourth downs yesterday. Fourth and 5 from the 41 is one where generalized analytics may say go for it but no way should applying one's brain to these specific belligerents/this situation - what is likely to happen with our offense going against that defense - lead one to think we should go for that fourth down at that time in a tie game. We deserved to get sacked. Fourth and 3 from the 44 is two yards less stupid but probably still not using your brain to account for opponent and it being a one score game. At what point does the riverboat gambler mentality cease being a ballsy way to show you play to win and start being bush league desperation. We're probably there. In many of the individual cases (including both above), one can credibly defend Taylor's analytics and/or circumstance-driven decisions, but in the aggregate we are the #73 fourth down team yet go for it #3 of any team in the country (Baylor and Air Force also go for it too much). That seems to leave points and field position on the table. I suspect we are a worse team than we would be if Taylor were more conservative.

As always, I enjoy your thoughtful examination of the previous game. Indeed this one must have been very discouraging when all aspects of the game were painfully poor. I would appreciate any "analytics" that you (or anybody else) have regarding the 4th down plays. I am familiar with the work done in the NFL and the papers published on that subject, but I am unsure of exactly what analytic "data" (if any) is driving TT's decisions. I totally agree with you that  going for it in the two cited cases was a very bad choice.

IMO there are all kinds of potential problems with using "statistics" to guide decisions in football, the main one being the applicability of aggregate population values to a specific population (Stanford's team) that is in no way "typical" of the general population. As you point out, we are unfortunately way out on the "tail" of the distribution in both offense and defense. What may "typically" work for a "middle of the pack" team probably will not for the bottom 1-2%.

Agree. Moreover, there's no reason why an analytics system can't be tailored to a specific matchup prior to or even during a game. I know nothing about being a football coach, but if it were my job, I'd have a really smart statistician in the booth feeding analytics and time management to me constantly. 

Somewhat off-topic - take a look at Kyle Smith's message and depth around analytics. It sounds like "Moneyball" for hoops - his staff tracks over 50 additional stats that are not normally tracked. (I'm assuming these include actions like defelections on defense and successful screens on offense.) These stats are categorized into winning and non-winning plays, and players receive very nuanced feedback on their performance at all times. It seems like a much better, more nuanced and more actionable version of the +/- system. Smith also has a lot to say about shot selection based on analytics. 

I'd love to know more about the analytical basis for Taylor's 4th down decisions. I suspect it's based on the generic studies and data, combined with a feeling that our chances of winning are extremely low either way, so we have to pull out all the stops. I do find it ironic that we've gone from one end of the spectrum (Shaw) to the other extreme end (TT) in terms of 4th down aggressiveness and reverence for field position. Between the two, I prefer TT's approach, but it all depends on what kind of team you have.

[CardinalSagehen]

6.  We went one of four on fourth downs yesterday. Fourth and 5 from the 41 is one where generalized analytics may say go for it but no way should applying one's brain to these specific belligerents/this situation - what is likely to happen with our offense going against that defense - lead one to think we should go for that fourth down at that time in a tie game. We deserved to get sacked. Fourth and 3 from the 44 is two yards less stupid but probably still not using your brain to account for opponent and it being a one score game. At what point does the riverboat gambler mentality cease being a ballsy way to show you play to win and start being bush league desperation. We're probably there. In many of the individual cases (including both above), one can credibly defend Taylor's analytics and/or circumstance-driven decisions, but in the aggregate we are the #73 fourth down team yet go for it #3 of any team in the country (Baylor and Air Force also go for it too much). That seems to leave points and field position on the table. I suspect we are a worse team than we would be if Taylor were more conservative.

As always, I enjoy your thoughtful examination of the previous game. Indeed this one must have been very discouraging when all aspects of the game were painfully poor. I would appreciate any "analytics" that you (or anybody else) have regarding the 4th down plays. I am familiar with the work done in the NFL and the papers published on that subject, but I am unsure of exactly what analytic "data" (if any) is driving TT's decisions. I totally agree with you that  going for it in the two cited cases was a very bad choice.

IMO there are all kinds of potential problems with using "statistics" to guide decisions in football, the main one being the applicability of aggregate population values to a specific population (Stanford's team) that is in no way "typical" of the general population. As you point out, we are unfortunately way out on the "tail" of the distribution in both offense and defense. What may "typically" work for a "middle of the pack" team probably will not for the bottom 1-2%.

Agree. Moreover, there's no reason why an analytics system can't be tailored to a specific matchup prior to or even during a game. I know nothing about being a football coach, but if it were my job, I'd have a really smart statistician in the booth feeding analytics and time management to me constantly. 

Somewhat off-topic - take a look at Kyle Smith's message and depth around analytics. It sounds like "Moneyball" for hoops - his staff tracks over 50 additional stats that are not normally tracked. (I'm assuming these include actions like defelections on defense and successful screens on offense.) These stats are categorized into winning and non-winning plays, and players receive very nuanced feedback on their performance at all times. It seems like a much better, more nuanced and more actionable version of the +/- system. Smith also has a lot to say about shot selection based on analytics. 

I'd love to know more about the analytical basis for Taylor's 4th down decisions. I suspect it's based on the generic studies and data, combined with a feeling that our chances of winning are extremely low either way, so we have to pull out all the stops. I do find it ironic that we've gone from one end of the spectrum (Shaw) to the other extreme end (TT) in terms of 4th down aggressiveness and reverence for field position. Between the two, I prefer TT's approach, but it all depends on what kind of team you have.