75 replies to this topic

[vegas55]

Using statistical analysis to compute a players worth/value is only one aspect of moneyball. It's an aspect that Polian says wont work in the NFL and I tend to agree. But the other aspect of moneyball is the use of statistics, and not gut feelings or traditions, in determining game strategy decisions. The Oakland A s used the thousands of hours poured into sabermetrics to determine game day decisions. For example, sabermetrics demonstrated conclusively that the use of the bunt and most related small ball tactics were terrible strategy. Sacrificing an out just to move a base runner up 90 feet is bad strategy, regardless of the experience and gut instincts of a manager. The A s exploited this, and in fact, bunting by a non pitcher has become fairly rare in MLB

So I hope the Bills embrace this aspect of moneyball in making statistically/scientifically based game management decisions. At the very least, we would see a lot less punts from the 34 yard line

[HamSandwhich]

vegas55, on 04 January 2013 - 03:22 PM, said:

Using statistical analysis to compute a players worth/value is only one aspect of moneyball. It's an aspect that Polian says wont work in the NFL and I tend to agree. But the other aspect of moneyball is the use of statistics, and not gut feelings or traditions, in determining game strategy decisions. The Oakland A s used the thousands of hours poured into sabermetrics to determine game day decisions. For example, sabermetrics demonstrated conclusively that the use of the bunt and most related small ball tactics were terrible strategy. Sacrificing an out just to move a base runner up 90 feet is bad strategy, regardless of the experience and gut instincts of a manager. The A s exploited this, and in fact, bunting by a non pitcher has become fairly rare in MLB

So I hope the Bills embrace this aspect of moneyball in making statistically/scientifically based game management decisions. At the very least, we would see a lot less punts from the 34 yard line

I'm excited, for one, to see if they apply this and how successful it is.  At least it's something new!

[K-9]

vegas55, on 04 January 2013 - 03:22 PM, said:

Using statistical analysis to compute a players worth/value is only one aspect of moneyball. It's an aspect that Polian says wont work in the NFL and I tend to agree. But the other aspect of moneyball is the use of statistics, and not gut feelings or traditions, in determining game strategy decisions. The Oakland A s used the thousands of hours poured into sabermetrics to determine game day decisions. For example, sabermetrics demonstrated conclusively that the use of the bunt and most related small ball tactics were terrible strategy. Sacrificing an out just to move a base runner up 90 feet is bad strategy, regardless of the experience and gut instincts of a manager. The A s exploited this, and in fact, bunting by a non pitcher has become fairly rare in MLB

So I hope the Bills embrace this aspect of moneyball in making statistically/scientifically based game management decisions. At the very least, we would see a lot less punts from the 34 yard line

Bunting and punting are VERY similar, indeed.

GO BILLS!!!

[Just in Atlanta]

I've said this many times here--the team that first cuts punts and FG attempts down dramatically--intelligently, not recklessly--will prosper. It would also be interesting to see more onside kicks. Read this stat many times, don't have a source, but the success rate of onside kicks in the 3rd quarter is about 60 percent.

I've never understood why coaches are so conservative. Maybe it will take moneyball and someone like Chip Kelly to turn the old, boring model upside down.

Edited by Just in Atlanta, 04 January 2013 - 03:47 PM.

[RealityCheck]

The physical universe and all it's subdivisions can be reduced to mathematical expressions. The concept is sound despite the obvious struggles to properly identify the actual variables at play and then the pain of creating formulas to plug your variables into. As complex as analyzing football appears to be it will eventually be broken down fairly accurately with the right algorithms. Granted it may take decades to completely hash out but it can be done. Far more complex systems than football have been broken down into fairly accurate tables of statistical probabilities. The one aspect of football that might be the most problematic is the fact that emotion is the ultimate X-factor. In many sports strong emotional states of mind lead to mistakes. In football, that aspect alone can overcome a variety of measurable deficiencies in your team.

[Chandemonium]

K-9, on 04 January 2013 - 03:24 PM, said:

Bunting and punting are VERY similar, indeed.

GO BILLS!!!

In fact, by my advanced analytical calculations, they differ by only one letter.

[NoSaint]

Just in Atlanta, on 04 January 2013 - 03:43 PM, said:

I've said this many times here--the team that first cuts punts and FG attempts down dramatically--intelligently, not recklessly--will prosper. It would also be interesting to see more onside kicks. Read this stat many times, don't have a source, but the success rate of onside kicks in the 3rd quarter is about 60 percent.

I've never understood why coaches are so conservative. Maybe it will take moneyball and someone like Chip Kelly to turn the old, boring model upside down.

well, that 40% fail rate is a tough pill to swallow when giving the ball nearly in scoring position.  (i know its statically not awful but it takes a lot to risk it)

and if a team is known to regularly do it, youd expect that success rate to go down. part of why they are so effective is because guys get lazy in that responsibility and then end up targeted by surprise.

its an interesting study though - and im not dismissing its potential effectiveness over the long run as i think coaches make a lot of safe choices that make their press conference easier and disregard that it makes the game harder to win.

Edited by NoSaint, 04 January 2013 - 04:47 PM.

[LittleJoeCartwright]

RealityCheck, on 04 January 2013 - 03:57 PM, said:

The physical universe and all it's subdivisions can be reduced to mathematical expressions. The concept is sound despite the obvious struggles to properly identify the actual variables at play and then the pain of creating formulas to plug your variables into. As complex as analyzing football appears to be it will eventually be broken down fairly accurately with the right algorithms. Granted it may take decades to completely hash out but it can be done. Far more complex systems than football have been broken down into fairly accurate tables of statistical probabilities. The one aspect of football that might be the most problematic is the fact that emotion is the ultimate X-factor. In many sports strong emotional states of mind lead to mistakes. In football, that aspect alone can overcome a variety of measurable deficiencies in your team.

Yeah, here's a good one:

The percentages of converted third-and-ones with a 5 WR set with empty backfield.

[K-9]

I'd be interested in knowing what percentage of the 40% of failed onside kicks in the third Qtr lead to points for the opposing team.

After that I'd want to know what percentage of opposing teams that didn't score were still able to flip the field in terms of field position.

Momentum is a fickle thing in football.

GO BILLS!!!

[NoSaint]

LittleJoeCartwright, on 04 January 2013 - 05:22 PM, said:

Yeah, here's a good one:

The percentages of converted third-and-ones with a 5 WR set with empty backfield.

its become incredibly common, especially with receivers protected. completing a short throw with the space created by 5 wide is mathematically approaching "fall forward for a foot" if you have even a halfway decent qb

[since79]

On the subject of punting and onside kicks.   Here is article of a. HS who did not punt and used 7 variations of the os kick to win a state title.

http://highschool.ri....asp?CID=892888

Some articles talking about punting.

http://www.newyorker...ot-to-punt.html

http://www.nytimes.c...ourth-down.html


Paper analyzing punting in NFL sited in articles.

http://emlab.berkele...B_CORRECTED.pdf


Go for it!

[playman]

if this helps to identify players who are more likely to perform then im all for it. i really don´t think " money " is key here but rather " production "

for example, are a QB´s stats all about him or the players around him.

i´m pretty sure no USC QB will pass this test

[Just in Atlanta]

NoSaint, on 04 January 2013 - 04:46 PM, said:

well, that 40% fail rate is a tough pill to swallow when giving the ball nearly in scoring position.  (i know its statically not awful but it takes a lot to risk it)

and if a team is known to regularly do it, youd expect that success rate to go down. part of why they are so effective is because guys get lazy in that responsibility and then end up targeted by surprise.

its an interesting study though - and im not dismissing its potential effectiveness over the long run as i think coaches make a lot of safe choices that make their press conference easier and disregard that it makes the game harder to win.

Going for it on 4th far more often is a better strategy. But onside kick increases should be looked at too, in conjunction. The combined effect would cause all sorts of stresses.

If teams are thinking they need to prepare for these, they are more apt to make a mistake when a normal punt happens.

Further, a 40 percent fail rate means we get it 60 percent of the time and that the opponent only gets it where the ball is recovered. All in all, that's not bad IMO.

[RealityCheck]

LittleJoeCartwright, on 04 January 2013 - 05:22 PM, said:

Yeah, here's a good one:

The percentages of converted third-and-ones with a 5 WR set with empty backfield.

Excellent question. You could look at simplistically in terms of particular QBs, but that isn't really an analysis. It would be more beneficial to not only take the conversion % but superimpose it over the % of various defensive formations and personnel packages by team and in which quarters vs. score differential and grade the quality of the play call based on who actually got open vs said coverages and if the QB went to the right guy or into the teeth of the coverage. This would allow you to properly grade the design of the play and both the decision making and accuracy of the QB. Most NFL defenses play man coverage and back the safeties off to protect against the deep ball in those situations unless they're within their own 20. Lots of DCs will call a coverage that is weak underneath so that if the catch is made then it's not for a big gain and not some 20 yarder. Personally I hate any 3rd and short package that doesn't keep a RB in the backfield to at least pick up a blitzer, delayed handoff, or an inside screen if you're not gonna run it. Chan was calling plays like he had Brees or Brady out there. I hope by now he is at least aware that it was just a guy named Fitzgerald.

[KRC]

I have seen it mentioned a few times in this thread. it is not sabremetrics. It is SABR-metrics, named after SABR. That is the Society for American Baseball Research. They are the baseball equivalent of my organization the PFRA (Professional Football Researchers Association).

Now, to the topic at hand. Can you come up with a way to use statistics to improve the team? I say it is possible. However, it is far more complex than in baseball. As mentioned previously, baseball lends itself to individual player statistics. Football does not. Therefore, you have to look at how individual player statistics can be applied to the team.

For example, I developed a rusher efficiency stat for Cold Hard Football Facts. It was very effective in determining how efficient your running game is in various situations. You can look at individual rushers to see if one is better in certain situations or you can look at it from a team perspective to see if there are weaknesses in your running game. This combines the running back, as well as the offensive line. When you analyze the numbers, you can see whether the running back is causing the deficiency or whether it is an offensive linemen. You can also break it down to see if it is that offensive lineman or running back against a specific defense, or in general. You can also normalize the stats to see if the running back is declining in efficiency over his career.

I think that similar philosophies can be implimented, as long as the team approach is used, with the ability to analyze the stat in the context of a specific player in various situations.

[VABills]

Are we doing Moneyball for the coach?  Who has greatest win-loss, what they call on given downs, peer reviews on how well they do, are liked?