Going for it and clock management

[GaryPinC]

He did not "fudge together his data". He explained exactly what information he based his study on. While you have pointed out some things he might have done differently, you don't have any evidence that by including the data you say he left out, the result would have been any different, or how it would have differed. I think you also mischaracterize his conclusions. The author never says "Inside your own 20, fourth and 2 you should always go for it." He makes it clear that decisions will (and should) frequently vary depending upon the score of the game and the amount of time remaining. There here have been multiple studies of these issues and to my knowledge all have concluded that NFL coaches punt far, far too often.

 

No, he explained in a relative way what he did. For example, he did not mention borrowing the 3rd down data until responding to a comment below the article.

 

If you have a link to where his actual study is it would be a great help if you would provide it. A peer reviewed statistical publication would be fantastic. I took a quick look but couldn't find it. You are correct, he did not explicitly say "you should always go for it" so my apologies for misspeaking. His graph, however, does recommend going for it inside your own 20 with fourth and 1, quite often fourth and two.

 

"He makes it clear that decisions will (and should) frequently vary depending upon the score of the game and the amount of time remaining." So maybe go for it, unless a coach in a game situation doesn't think it's wise. Fantastic stuff.

 

Without his actual study (only him presenting graphs in an article on his website) he (or I) may be mischaracterizing the conclusions.

 

This I do know: You don't just throw out half a football game because you think the time during the game is inappropriate. You prove it with statistical significance not a subjective, unproven statement. Common sense hypotheses or arguments need to be statistically justified to produce a strong research paper.

 

How about how he chooses to define "expected points"?: "EP is the average potential points a team can expect given a certain situation. The most common example is the potential point value of a 1st down at each yard line on the field. EP is the average of all 'next' score values at any given yard line. It's not necessary the average points scored on the current possession because possession could be exchanged several times before the 'next' score. EP is positive when the offense will usually score next, and negative if the defense will usually score next."

 

 

So if the Bills get the ball at their nine, punt, the other team fumbles or has a bunch of penalties and we end up getting the ball at their 15 and punch in the touch down, by Brian's definition those points apply to when the Bills started at their own 9.

 

Is that really valid? Or is it more a function of starting at opponents 15? If it is not valid at the Bills 9 yard line then the expected point value would become more negative on the 9 and skew his fourth down recommendations.

 

I don't know the answers. All I am pointing out is the dogma level is running huge on this right now. I think most any intelligent football fan suspects that coaches are too conservative on certain fourth downs and Brian certainly does a great job of subjectively pandering to that school of thought. I would like to see his methodologies and be wrong about the subjectivity part. But researchers who use methodical, statistically proven criteria for selecting data and defining parameters don't typically fall back on subjective justifications . They clearly state their proof.

Edited May 9, 2013 by GaryPinC

[eball]

How about how he chooses to define "expected points"?: "EP is the average potential points a team can expect given a certain situation. The most common example is the potential point value of a 1st down at each yard line on the field. EP is the average of all 'next' score values at any given yard line. It's not necessary the average points scored on the current possession because possession could be exchanged several times before the 'next' score. EP is positive when the offense will usually score next, and negative if the defense will usually score next."

 

So if the Bills get the ball at their nine, punt, the other team fumbles or has a bunch of penalties and we end up getting the ball at their 15 and punch in the touch down, by Brian's definition those points apply to when the Bills started at their own 9.

 

Is that really valid? Or is it more a function of starting at opponents 15? If it is not valid at the Bills 9 yard line then the expected point value would become more negative on the 9 and skew his fourth down recommendations.

 

I think you're misinterpreting his definition of EP value. While EP is "not necessarily" the average pts scored on the current possession, points expected from the current possession is still what it represents. So in your example, I believe the author would agree with you that the EP value is for the Bills starting at the opponent's 15.

 

Or I could be completely wrong.

[reddogblitz]

Stats are great, but if Pete Carrol had taken the gimme FG at the end of the 1st half vs Atlanta instead of being a hot dog and going for it, his team would probably have been in the NFC Championship game.

[mannc]

 

You are correct, he did not explicitly say "you should always go for it" so my apologies for misspeaking. His graph, however, does recommend going for it inside your own 20 with fourth and 1, quite often fourth and two.

 

"He makes it clear that decisions will (and should) frequently vary depending upon the score of the game and the amount of time remaining." So maybe go for it, unless a coach in a game situation doesn't think it's wise. Fantastic stuff.

 

 

Yes, I think he does say it is usually a good idea to go for it on fourth and 1 from your own 20 and I think he explains pretty convincingly why that's the case. You seem to be taking issue with that because it is so entirely contrary to what every coach in the NFL has ever done and, perhaps equally importantly, what every announcer says the coach should do. But that's precisely the point: the conventional wisdom is wrong and the decisions most coaches make wihout even thinking about it, like punting from your own 20 on fourth and 1, can't be statistically justified. The fact that he's not able to categorically say "Thou shalt always punt on 4th and less than 2 from your own 20" does not weaken his overall point. There are situations where pinning your opponent back 35 more yards and making them go farther to score can essentially end the game. But that's usually not going to be the case and it will be up to the coaches to understand when that is.

 

Look, his study isn't perfect, and football is very hard to quantify because there are an infinte number of variables, unlike baseball. But at least he's making a very good attempt to quantify decisions that have for decades simply been made "by gut", without any real thought about actual statistical probabilities.

[eball]

Stats are great, but if Pete Carrol had taken the gimme FG at the end of the 1st half vs Atlanta instead of being a hot dog and going for it, his team would probably have been in the NFC Championship game.

 

But that point only emphasizes what the study author is saying...going for it doesn't guarantee success, but the fear of failure is what more often than not sways coaches to make the safe call -- even if it isn't with the percentages. You're criticizing Carroll for taking the percentage play and failing, which is what all fans and media do in these situations because they only look at the specific play and not the game as a whole.

 

Basically, you need a coach with the balls to stand up there after the game and say, "it was the percentage play, it didn't work, but I'm always going to go with what I believe gives us the best chance."

Edited May 9, 2013 by eball

[mannc]

But that point only emphasizes what the study author is saying...going for it doesn't guarantee success, but the fear of failure is what more often than not sways coaches to make the safe call -- even if it isn't with the percentages. You're criticizing Carroll for taking the percentage play and failing, which is what all fans and media do in these situations because they only look at the specific play and not the game as a whole.

 

Basically, you need a coach with the balls to stand up there after the game and say, "it was the percentage play, it didn't work, but I'm always going to go with what I believe gives us the best chance."

Exactly correct. Another great example was the Pats-Colts game a couple years ago, when Belichick went for it on 4th and 2 from his own 34, with the lead. He was playing the percentages, but got crucified in the media for making a crazy gamble. If he had punted, and the Colts had marched the length of the field to score, no one would ever have questioned it.

[GaryPinC]

I think you're misinterpreting his definition of EP value. While EP is "not necessarily" the average pts scored on the current possession, points expected from the current possession is still what it represents. So in your example, I believe the author would agree with you that the EP value is for the Bills starting at the opponent's 15.

 

Or I could be completely wrong.

 

Thanks, good thought. Maybe I am wrong. My understanding is that expected points at any given yardline is based on data of who scores "next" If the Offense, it's positive. If the defense gets it and scores it's negative.

 

EP is the average of all 'next' score values at any given yard line. It's not necessary the average points scored on the current possession because possession could be exchanged several times before the 'next' score.

 

I interpret this to say that he uses next score (by either side) to determine the EP and that yardline. The underlined portion led me to believe that even if no one scores on that possession he would attribute the very next score to that original yard line. Let me know what you think.

 

 

Yes, I think he does say it is usually a good idea to go for it on fourth and 1 from your own 20 and I think he explains pretty convincingly why that's the case. You seem to be taking issue with that because it is so entirely contrary to what every coach in the NFL has ever done and, perhaps equally importantly, what every announcer says the coach should do. But that's precisely the point: the conventional wisdom is wrong and the decisions most coaches make wihout even thinking about it, like punting from your own 20 on fourth and 1, can't be statistically justified. The fact that he's not able to categorically say "Thou shalt always punt on 4th and less than 2 from your own 20" does not weaken his overall point. There are situations where pinning your opponent back 35 more yards and making them go farther to score can essentially end the game. But that's usually not going to be the case and it will be up to the coaches to understand when that is.

 

Look, his study isn't perfect, and football is very hard to quantify because there are an infinte number of variables, unlike baseball. But at least he's making a very good attempt to quantify decisions that have for decades simply been made "by gut", without any real thought about actual statistical probabilities.

 

Actually my point is that it isn't a very good study and I don't think his data makes a strong enough case for a coach to really trust it. I don't see any error bars associated with this study. The author processed a lot of numerical data which has error associated with it. When he says that EP is 0.3, 0.4 but going for it has an EP of 1.2 which he feels is clearly better we don't know that without error and statistical analysis. The error could easily be so great that there is no difference in any of those three numbers. Maybe his results/conclusions would hold up but the study is really weak imo.

Edited May 9, 2013 by GaryPinC

[The Wiz]

Right now I'm kicking myself for not asking this question during yesterday's ST holder call w/ Marrone. I'd love to know his thoughts on using statistics and probabilities to make 4th down decisions, as opposed to taking the "safe" route like so many coaches do.

I thought that was part of the Bills new analytic department that still doesn't exist yet?

 

I get the feeling that with him making his NFL debut we will probably see a lot of bread and butter (safe) play calling. If the team gets in a groove (good or bad) I would imagine he would loosen the belt and start throwing in more and more trick plays.

[BuffaloBill]

The idea that free agents are overvalued and draft choices. Are undervalued flies in the face of fan pressure. Fans want the hype of FA's but the reality is that a successful draft will do more for a team.

[Joe W]

Read the grant land article about the Cleveland Browns and how they're serious about analytics. Branden, the tool that he is, has yet to follow up on any of his early January grandiose promises. The guy is inept

[eball]

Thanks, good thought. Maybe I am wrong. My understanding is that expected points at any given yardline is based on data of who scores "next" If the Offense, it's positive. If the defense gets it and scores it's negative.

 

EP is the average of all 'next' score values at any given yard line. It's not necessary the average points scored on the current possession because possession could be exchanged several times before the 'next' score.

 

I interpret this to say that he uses next score (by either side) to determine the EP and that yardline. The underlined portion led me to believe that even if no one scores on that possession he would attribute the very next score to that original yard line. Let me know what you think.

 

I'm confused as hell.

 

Actually, I still think what I wrote above. The "changing possessions" factor is considered when he's deriving the overall average that leads to the EP value for each yardline. A team's EP constantly changes. Going back to your example, the Bills would have one EP for the possession that started at the 9, and then a new EP for the possession that starts at the opponent's 15.

 

Or not.

[mannc]

Read the grant land article about the Cleveland Browns and how they're serious about analytics. Branden, the tool that he is, has yet to follow up on any of his early January grandiose promises. The guy is inept

I don't know. I did not read anything in the Grantland article that convinced me that the "new" Browns organization has any better idea what they are doing than the previous regime. It actually made them seem like a bunch of tools. We'll see, I guess.

[GaryPinC]

I'm confused as hell.

 

Actually, I still think what I wrote above. The "changing possessions" factor is considered when he's deriving the overall average that leads to the EP value for each yardline. A team's EP constantly changes. Going back to your example, the Bills would have one EP for the possession that started at the 9, and then a new EP for the possession that starts at the opponent's 15.

 

Or not.

 

Well, good. We're in the same boat of confusion. Ok, I hear what you're saying. Can we agree that the score from the opponents 15 would be factored into the EP (in some way) at the 9? It is, by the author's own definition, the "next" score.

 

From reading the author's articles, he needs the EP value at each yardline for establishing the validity of punting vs going for it vs field goal. Compare the EP values of a net 37 yard punt vs EP value at the first down mark relative to EP at your current position.

 

And I assume he uses "next" score to lend some weight to the team winning the field position battle. Having the ball inside your 10 means if you fail to move the ball the other team will more likely score in their next possession or two. But your team will also have another possession and once the next score occurs do you attribute it to each and every failed drive both teams had?

 

This gets at my point for the quality of the study. I think his own generic definition of "next points" speaks to the crappy quality of his research. He could easily establish and validate statistical criteria for which drives "next" scores are valid. For instance, he should be able to establish that if your field position on your next drive changes by more than 20 yds compared to previous, attributing "next" score would only be valid on the new field position.

 

This would be easy to establish, explain and include in the article. But he doesn't.

Edited May 10, 2013 by GaryPinC

[Just in Atlanta]

Stats are great, but if Pete Carrol had taken the gimme FG at the end of the 1st half vs Atlanta instead of being a hot dog and going for it, his team would probably have been in the NFC Championship game.

That's a statement in a vacuum. I don't remember that call specifically, but who's to say they weren't put in the position of being so close because of going for it on another 4th down? Whose to say they didn't win two games for going for it more often on 4th?

 

I'm not saying you're wrong, or right. But your statement solidifies why coaches are so ultra, ultra conservative. Fans look at one instance in a vacuum, and forget the body of work. All it takes is one loss and many fans forget the many wins.

[reddogblitz]

Really there are only 2 decisions a coach can make in such a situation. The smart one and the dumb one. The smart one is one that works while a dumb one does not. If you go for it on 4th down and get it, you are a coaching genius. If you don't get it, you're an idiot.

 

I don't necessarily subscribe to the idea that coaches make the decisions they do because they're afraid to fail. I think they make the decisions they do thinking it's best chance to win. Kicking a FG may not be sexy and daring, but it's points. After a few, FGs start adding up to real points.

 

I want my coach to kick FGs on 4th down and punting on 4th down in most cases.

 

If a coach does go whole hog in on this kindof theory it would be interesting. I question if it would really be successful or not. Perhaps we'll find out.

[GaryPinC]

FWIW: Here is an article discussing the statistics and rationale for Chip Kelly going for it on fourth down at Oregon:

 

http://articles.philly.com/2013-03-27/sports/38043609_1_de-anthony-thomas-desean-jackson-s-wide-receiver

 

"Punt 4th down? Hell, yeah

In case you were wondering, Chip Kelly does plan on carrying a punter on his 53-man roster. Might not have a fullback, but he'll have a punter. Those rumors that he always went for it on fourth down at Oregon are slightly exaggerated.

 

The Ducks went for it on fourth down 31 times last season (and converted 20 of them). That total was only 10th among the NCAA's 120 Football Bowl Subdivision members.

 

During Kelly's four seasons as Oregon's head coach, the Ducks averaged just 2.2 fourth-down attempts per game. They finished tied for 17th in the FBS in fourth-down attempts in 2009 (22), second in '10 (34) and fifth in '11 (31).

 

In the Brooks Brothers NFL, usually only bad teams go for it on fourth down. Last year, only four teams had more than 18 fourth-down tries: 2-14 Jacksonville (26), 5-11 Arizona (24), 7-8-1 St. Louis (24) and your 4-12 Eagles (24).

 

Asked about his reputation as a fourth-down gambler last week, Kelly said: "I think there's fallacy and reality. I don't think very often we went for it on fourth down on our side of the field. It would be once or twice a season, depending on the situation."

 

He pointed out that the leg strength of his kicker often factors into fourth-down decisions.

 

"If you don't have a guy that can kick a long field goal, what are you going to do when the ball is on the 37-yard line?" he said. "Will you kick a 52-yarder or are you going to punt it? If [the punt] goes into the end zone, you have a net of 17 yards. Or do you go for it because you have a good defense and you're not averse to putting them on the field on the 37-yard line?"

 

Kelly didn't have a kicker with a big leg at Oregon. In his four seasons as head coach, the team's longest field goal was 43 yards. The last 3 years, the Ducks were just 5-for-14 on field goal attempts from 40-plus yards."

 

Doesn't sound so radical.

[mannc]

FWIW: Here is an article discussing the statistics and rationale for Chip Kelly going for it on fourth down at Oregon:

 

http://articles.phil...s-wide-receiver

 

"Punt 4th down? Hell, yeah

In case you were wondering, Chip Kelly does plan on carrying a punter on his 53-man roster. Might not have a fullback, but he'll have a punter. Those rumors that he always went for it on fourth down at Oregon are slightly exaggerated.

 

The Ducks went for it on fourth down 31 times last season (and converted 20 of them). That total was only 10th among the NCAA's 120 Football Bowl Subdivision members.

 

During Kelly's four seasons as Oregon's head coach, the Ducks averaged just 2.2 fourth-down attempts per game. They finished tied for 17th in the FBS in fourth-down attempts in 2009 (22), second in '10 (34) and fifth in '11 (31).

 

In the Brooks Brothers NFL, usually only bad teams go for it on fourth down. Last year, only four teams had more than 18 fourth-down tries: 2-14 Jacksonville (26), 5-11 Arizona (24), 7-8-1 St. Louis (24) and your 4-12 Eagles (24).

 

Asked about his reputation as a fourth-down gambler last week, Kelly said: "I think there's fallacy and reality. I don't think very often we went for it on fourth down on our side of the field. It would be once or twice a season, depending on the situation."

 

He pointed out that the leg strength of his kicker often factors into fourth-down decisions.

 

"If you don't have a guy that can kick a long field goal, what are you going to do when the ball is on the 37-yard line?" he said. "Will you kick a 52-yarder or are you going to punt it? If [the punt] goes into the end zone, you have a net of 17 yards. Or do you go for it because you have a good defense and you're not averse to putting them on the field on the 37-yard line?"

 

Kelly didn't have a kicker with a big leg at Oregon. In his four seasons as head coach, the team's longest field goal was 43 yards. The last 3 years, the Ducks were just 5-for-14 on field goal attempts from 40-plus yards."

 

Doesn't sound so radical.

Good stuff. Thanks. It's true that Oregon had lousy placekickers, which to some extent explains the high number of 4th down attempts. (They actually have had pretty good punters, however.) But it is also worth remembering that the majority of the Ducks' games the past 3 or 4 years were massive blowouts. It is safe to assume that when you are up by 40 points in the second half, you are not going to go for it on fourth down, no matter what the distance is. Having watched almost every Ducks' game in the Kelly era, I can tell you Kelly is far more inclined to go for it on 4th down in meaningful situations than any NFL coach, with the possible exception of Belichick, and every major college coach I know of.