Monday Morning Robocoach
December 9, 2013 6:20 AM Subscribe
Watching one of the exciting snow-bound football games yesterday, the thought may have occurred to you: If I was a coach, would I go for it on this 4th down? This bot from the New York Times will tell you, and maybe even add a little attitude to the answer, which is usually much more aggressive than NFL coaches.
That Eagles-Lions game was amazing. I'm not sure I've ever seen a game with so much snow on the field. That shot of Calvin Johnson with his helmet packed full of snow! No discussion of snow football is complete without reference to Mark Henderson, however.
posted by Rock Steady at 6:36 AM on December 9, 2013 [7 favorites]
posted by Rock Steady at 6:36 AM on December 9, 2013 [7 favorites]
That bot is cute but I was hoping for more of a football Krang.
posted by troika at 6:39 AM on December 9, 2013 [2 favorites]
posted by troika at 6:39 AM on December 9, 2013 [2 favorites]
To be fair, Neville Chamberlain can be said to have been "much more aggressive than NFL coaches."
Is there a breakdown of which coaches/teams match up with the bot the most? It'd be interesting to get the coaches take on the bot and why they might have made a different call in the moment.
The bot is sparkly-new this week, but I'm sure the NFL's Second Season (February-August) will feature plenty of opportunities for some SportsCenter interviews on that subject, at which point I predict every single coach will respond with, "Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field" or some such nonsense.
I was most interested to see how often a team shouldn't bother with close field goals -- essentially, any 4th-and-goal situation is effectively a net loss of points if you kick it.
posted by Etrigan at 6:39 AM on December 9, 2013
Is there a breakdown of which coaches/teams match up with the bot the most? It'd be interesting to get the coaches take on the bot and why they might have made a different call in the moment.
The bot is sparkly-new this week, but I'm sure the NFL's Second Season (February-August) will feature plenty of opportunities for some SportsCenter interviews on that subject, at which point I predict every single coach will respond with, "Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field" or some such nonsense.
I was most interested to see how often a team shouldn't bother with close field goals -- essentially, any 4th-and-goal situation is effectively a net loss of points if you kick it.
posted by Etrigan at 6:39 AM on December 9, 2013
snarfles: Is there a breakdown of which coaches/teams match up with the bot the most? It'd be interesting to get the coaches take on the bot and why they might have made a different call in the moment.
Bill Barnwell's weekly "Thank You For Not Coaching" column is a fascinating breakdown of coaching decisions that focuses on decision-making and not on outcome.
posted by Rock Steady at 6:48 AM on December 9, 2013 [4 favorites]
Bill Barnwell's weekly "Thank You For Not Coaching" column is a fascinating breakdown of coaching decisions that focuses on decision-making and not on outcome.
posted by Rock Steady at 6:48 AM on December 9, 2013 [4 favorites]
The robocoach would go for it on 4th and 3 FROM ITS OWN 8 YARD LINE.
posted by Ghost Mode at 6:51 AM on December 9, 2013 [5 favorites]
posted by Ghost Mode at 6:51 AM on December 9, 2013 [5 favorites]
This is fascinating. A few weeks ago, Hang Up and Listen profiled a high school coach who never punts. If going for it didn't put the coach directly in the hot seat, I think the NFL would experiment a lot more.
posted by gladly at 6:51 AM on December 9, 2013 [4 favorites]
posted by gladly at 6:51 AM on December 9, 2013 [4 favorites]
We've already seen it this year. There have been vastly more attempts at 4th and 3 or less, and most of them succeeding. It used to be that a 4th and 3 at midfield was a stop, now, it's no such thing.
It's what let the Bears beat the Packers, when they went for a 4th and 1, converted, then went on a 9 minute drive to score a FG and kill the game by leaving no clock, and it's what's killed the Bears otherwise, as teams have realized that not only should they be going for it more, they should be going for it even more against the Bear's devastated-by-injury defense.
Another big factor -- you go for 4th and goal on the 1 and get stuffed. Sucks, you walked away from three points, right? Well, the other team now has the ball on their 1. Get a stop, and you're right back on the attack -- and when you're running plays from your own end zone, that's prime pick-6 country.
The one time when that chart will turn wrong is when a FG turns a 2 score game into a 1 score game. In the last ten minutes, when they're going off of win percentage, they'll do much better, but if it's 4th and 3 at the 25 and I'm down 9 with 6 minutes left, I really want to make it down 6, aim for a stop, and then have a shot at the lead. But that situation is not as common as people think, and the win percentages, as opposed to just the play odds, will account for that.
In fact, in almost all cases, you will, over iterations, lose points for kicking a FG on a 4th and goal when you're on the 2 or 1. A TD is worth 6+ points, depending on how reliable your kicker is and how good at two point conversions you are, a FG is worth 3. If you can score the TD more than 47ish% of the time, you're leaving points on the ground if you kick -- exactly what percentage depends on how good at PATs and 2PATs you are.
Of course, you have to deal with the fan base calling you the WORST COACH EVAR when you go for it on a 4th and goal on the 1 and get stuffed. I vividly remember the "Bet the Bears wish they had the three points!!!" once this season this happened. I had to say "You mean minus three points? Because if they kicked the FG, they wouldn't have gotten the Pick 6 that happened the very next play!" See? They lost that one bet, but if they'd kicked, they wouldn't have been in the position where an interception would have a very high chance of a defensive score. In terms of one play, they walked away with 0 points. Over two plays, they walked away with 6+PAT. If they'd kicked the FG, the best they could do is 3, the worst? -3-PAT (if they run back the kickoff for a TD) Given the Bears defensive history of scoring points, putting them on the 1 yard line isn't the worst outcome in the world.
Remember: If you make the statistically right call, you can still lose *on that call.* It doesn't make it any less right, it just means you lost the bet that time.
On Preview: Etrigan notes leaving points on the ground on FGs on the 1 yard line, and Rock Steady saves me from having to dig out Bill Barnwell's "Thank You For Not Coaching" link. Thanks!
posted by eriko at 6:56 AM on December 9, 2013 [11 favorites]
It's what let the Bears beat the Packers, when they went for a 4th and 1, converted, then went on a 9 minute drive to score a FG and kill the game by leaving no clock, and it's what's killed the Bears otherwise, as teams have realized that not only should they be going for it more, they should be going for it even more against the Bear's devastated-by-injury defense.
Another big factor -- you go for 4th and goal on the 1 and get stuffed. Sucks, you walked away from three points, right? Well, the other team now has the ball on their 1. Get a stop, and you're right back on the attack -- and when you're running plays from your own end zone, that's prime pick-6 country.
The one time when that chart will turn wrong is when a FG turns a 2 score game into a 1 score game. In the last ten minutes, when they're going off of win percentage, they'll do much better, but if it's 4th and 3 at the 25 and I'm down 9 with 6 minutes left, I really want to make it down 6, aim for a stop, and then have a shot at the lead. But that situation is not as common as people think, and the win percentages, as opposed to just the play odds, will account for that.
In fact, in almost all cases, you will, over iterations, lose points for kicking a FG on a 4th and goal when you're on the 2 or 1. A TD is worth 6+ points, depending on how reliable your kicker is and how good at two point conversions you are, a FG is worth 3. If you can score the TD more than 47ish% of the time, you're leaving points on the ground if you kick -- exactly what percentage depends on how good at PATs and 2PATs you are.
Of course, you have to deal with the fan base calling you the WORST COACH EVAR when you go for it on a 4th and goal on the 1 and get stuffed. I vividly remember the "Bet the Bears wish they had the three points!!!" once this season this happened. I had to say "You mean minus three points? Because if they kicked the FG, they wouldn't have gotten the Pick 6 that happened the very next play!" See? They lost that one bet, but if they'd kicked, they wouldn't have been in the position where an interception would have a very high chance of a defensive score. In terms of one play, they walked away with 0 points. Over two plays, they walked away with 6+PAT. If they'd kicked the FG, the best they could do is 3, the worst? -3-PAT (if they run back the kickoff for a TD) Given the Bears defensive history of scoring points, putting them on the 1 yard line isn't the worst outcome in the world.
Remember: If you make the statistically right call, you can still lose *on that call.* It doesn't make it any less right, it just means you lost the bet that time.
On Preview: Etrigan notes leaving points on the ground on FGs on the 1 yard line, and Rock Steady saves me from having to dig out Bill Barnwell's "Thank You For Not Coaching" link. Thanks!
posted by eriko at 6:56 AM on December 9, 2013 [11 favorites]
It's easy for Robocoach to go for it on 4th down because Robocoach doesn't have to deal with the mainstream sports media. I mean, Bill Belichick is about as "untouchable" coach as there can be in this era and he was widely eviscerated for 4th and 2 (which was absolutely the right call).
posted by nathancaswell at 6:57 AM on December 9, 2013
posted by nathancaswell at 6:57 AM on December 9, 2013
So basically punting on 4th down is the buying IBM of football?
posted by Pope Guilty at 7:08 AM on December 9, 2013 [2 favorites]
posted by Pope Guilty at 7:08 AM on December 9, 2013 [2 favorites]
Well punting on 4th and 3 from the other team's territory. I don't see the point in ever punting in that situation unless you're up by 2 with a minute to go and you have a guy who can accurately put the ball out of play inside the 5 yard line. EDIT: Response to Pope Guilty.
posted by Mister_A at 7:19 AM on December 9, 2013
posted by Mister_A at 7:19 AM on December 9, 2013
The thing the robocoach doesnt take into account is the psychological effect some of these decisions can have. If you go for it on 4th and 1 from your own 10 and fail, and 5 seconds later the other team scores, well, the team is going to question your judgment as a coach. It's a bit deflating really.
posted by Mister_A at 7:22 AM on December 9, 2013
posted by Mister_A at 7:22 AM on December 9, 2013
Psychology is for the Weak Humons Not for Football.
When the Game is played by only Cletuses there will be Purity finally in the Sport.
posted by Potomac Avenue at 7:25 AM on December 9, 2013
When the Game is played by only Cletuses there will be Purity finally in the Sport.
posted by Potomac Avenue at 7:25 AM on December 9, 2013
Aaaaand Rob Gronkowski's career is basically over. ACL tear with MCL damage to a guy who required 4(!) surgeries for a broken forearm last year. It's amazing how quickly the Patriots went from the best Tight End corps in the history of the NFL to the Hoomanawanui Era.
posted by Rock Steady at 7:30 AM on December 9, 2013
posted by Rock Steady at 7:30 AM on December 9, 2013
The thing the robocoach doesnt take into account is the psychological effect some of these decisions can have. If you go for it on 4th and 1 from your own 10 and fail, and 5 seconds later the other team scores, well, the team is going to question your judgment as a coach. It's a bit deflating really.
On the other hand, the psychological effect of going for it and making it is said to be quite inflating.
But then, I believe that the ideas of "psychology" and "momentum" are vastly overstated in football. I wish I had a nickel for every time I've heard "That first down was a huge momentum shift in this game!" followed by "Oh, but now they've fumbled the ball... WHICH IS A HUGE MOMENTUM SHIFT!" Well, duh, jackass -- having the ball is better than not having the ball.
posted by Etrigan at 7:30 AM on December 9, 2013 [2 favorites]
On the other hand, the psychological effect of going for it and making it is said to be quite inflating.
But then, I believe that the ideas of "psychology" and "momentum" are vastly overstated in football. I wish I had a nickel for every time I've heard "That first down was a huge momentum shift in this game!" followed by "Oh, but now they've fumbled the ball... WHICH IS A HUGE MOMENTUM SHIFT!" Well, duh, jackass -- having the ball is better than not having the ball.
posted by Etrigan at 7:30 AM on December 9, 2013 [2 favorites]
This discussion really isn't complete without Gregg Easterbrook's punting guidelines:
It seems like a good enough set of guidelines to me.
posted by graymouser at 7:32 AM on December 9, 2013 [1 favorite]
- Inside your own 20, punt.
- From your 21 to 35, go for it on fourth-and-2 or less.
- From your 36 to midfield, go for it on fourth-and-3 or less.
- From the opposition 49 to opposition 30, go for it on fourth-and-4 or less.
- From the opposition 29 to opposition 3, go for it on fourth-and-3 or less.
- From the opposition 2 or 1, go for it.
- Exception: inside the opponent's 25, attempt a field goal if it's the fourth quarter and a field goal causes a tie or gives you the lead.
It seems like a good enough set of guidelines to me.
posted by graymouser at 7:32 AM on December 9, 2013 [1 favorite]
- Exception 2: if you are a Manning, go for it regardless, because you'll either make it anyway or entertain the crowds with Manningface if you don't.
I'd amend Greggggg's last guideline as eriko suggests: If it is late enough in the game, you may want to kick inside the opponent's 25 if the field goal cuts it from a 7+ point lead to a <7 point lead, but it shouldn't be a no-brainer, as it seems it is now.
posted by Rock Steady at 7:39 AM on December 9, 2013
posted by Rock Steady at 7:39 AM on December 9, 2013
Yesterday was an especially nutty day in the NFL. Matt Prater broke the longstanding record for longest field goal with a 64 yard cruise missile, LeSean McCoy broke the Eagles single-game rushing record in several inches of snow, and then there was the insane finish in Baltimore with five touchdowns in just a little over two minutes. I'd love to see the advice from a coaching bot trained only on the results of yesterday's games.
posted by tonycpsu at 7:40 AM on December 9, 2013 [3 favorites]
posted by tonycpsu at 7:40 AM on December 9, 2013 [3 favorites]
Oh gods, it would be like the Star Trek scene where Kirk tells the computer that he's lying.
posted by Etrigan at 7:42 AM on December 9, 2013 [3 favorites]
posted by Etrigan at 7:42 AM on December 9, 2013 [3 favorites]
LeSean McCoy broke the Eagles single-game rushing record in several inches of snow
The snow helped. (In the slippery, the ball carrier can make the cut easier than the defender can react to it.)
posted by notyou at 7:49 AM on December 9, 2013
The snow helped. (In the slippery, the ball carrier can make the cut easier than the defender can react to it.)
posted by notyou at 7:49 AM on December 9, 2013
notyou: " The snow helped. (In the slippery, the ball carrier can make the cut easier than the defender can react to it.)"
No doubt, but it was still fun to watch.
posted by tonycpsu at 7:51 AM on December 9, 2013 [2 favorites]
No doubt, but it was still fun to watch.
posted by tonycpsu at 7:51 AM on December 9, 2013 [2 favorites]
They can put an asterisk next to his name in the record books. It'll look like a snowflake!
posted by tonycpsu at 7:51 AM on December 9, 2013 [3 favorites]
posted by tonycpsu at 7:51 AM on December 9, 2013 [3 favorites]
Here's the 4th down plays with 1-3 yards to go on the opponent's side of the field for 2013, and the same for 2012. The numbers aren't hugely different as far as going for it vs. not, though you could probably break it down further into close games where going for it makes a difference & games where it's, say, more than a 14-point score differential & coaches are being extra conservative.
posted by zempf at 7:52 AM on December 9, 2013 [1 favorite]
posted by zempf at 7:52 AM on December 9, 2013 [1 favorite]
So basically punting on 4th down is the buying IBM of football?
Yeah, although I think there's a psychological effect that may be as important as the "you don't get fired for hiring IBM/you don't get fired for punting" impact, which is coaches really hate being "out of the game," and they are often willing to sacrifice an amazing amount of expected wins to avoid that situation. You see this even more starkly in decisions about whether to go for two after a touchdown, where many coaches absolutely refuse to go for it early even if they know they will need a conversion eventually (and knowing whether or not you make it impacts your strategy going forward).
posted by dsfan at 7:53 AM on December 9, 2013
Yeah, although I think there's a psychological effect that may be as important as the "you don't get fired for hiring IBM/you don't get fired for punting" impact, which is coaches really hate being "out of the game," and they are often willing to sacrifice an amazing amount of expected wins to avoid that situation. You see this even more starkly in decisions about whether to go for two after a touchdown, where many coaches absolutely refuse to go for it early even if they know they will need a conversion eventually (and knowing whether or not you make it impacts your strategy going forward).
posted by dsfan at 7:53 AM on December 9, 2013
Exception 3 - Don't count out Touchdown Tom
posted by Potomac Avenue at 7:53 AM on December 9, 2013
posted by Potomac Avenue at 7:53 AM on December 9, 2013
Agents attempt to optimize to multiple equillibria, news at 11.
posted by downing street memo at 7:55 AM on December 9, 2013
posted by downing street memo at 7:55 AM on December 9, 2013
The snow helped. (In the slippery, the ball carrier can make the cut easier than the defender can react to it.)
Somebody hasn't watched Shady this year.
posted by cashman at 8:00 AM on December 9, 2013 [1 favorite]
Somebody hasn't watched Shady this year.
posted by cashman at 8:00 AM on December 9, 2013 [1 favorite]
Just get rid of that superfluous 4th down. It's like football's appendix.
posted by clvrmnky at 8:02 AM on December 9, 2013
posted by clvrmnky at 8:02 AM on December 9, 2013
cashman: " Somebody hasn't watched Shady this year."
I've watched every Eagles game, including preseason, and I agree that the snow helped make him even more elusive yesterday. He's made a ton of guys miss this year, but he definitely used the conditions to his advantage against Detroit.
posted by tonycpsu at 8:02 AM on December 9, 2013
I've watched every Eagles game, including preseason, and I agree that the snow helped make him even more elusive yesterday. He's made a ton of guys miss this year, but he definitely used the conditions to his advantage against Detroit.
posted by tonycpsu at 8:02 AM on December 9, 2013
Exception 3 - Don't count out Touchdown Tom
And his trusty sidekick, Questionable Pass Interference Pete.
posted by inigo2 at 8:04 AM on December 9, 2013 [8 favorites]
And his trusty sidekick, Questionable Pass Interference Pete.
posted by inigo2 at 8:04 AM on December 9, 2013 [8 favorites]
I don't think so at all. Every move he made yesterday he did in regular conditions, and shook guys out of their shoes. Think about that Panthers preseason game. The conditions make it harder for the runner to cut as well, don't forget that.
posted by cashman at 8:04 AM on December 9, 2013
posted by cashman at 8:04 AM on December 9, 2013
This is really cool, but the model is clearly broken for 4th downs when backed up against your own goal line. Coaches might be too conservative in general, but this bot is too aggressive in general.
posted by grog at 8:15 AM on December 9, 2013 [2 favorites]
posted by grog at 8:15 AM on December 9, 2013 [2 favorites]
the model is clearly broken for 4th downs when backed up against your own goal line
Can you explain why you think this is the case?
posted by Potomac Avenue at 8:20 AM on December 9, 2013
Can you explain why you think this is the case?
posted by Potomac Avenue at 8:20 AM on December 9, 2013
the model is clearly broken for 4th downs when backed up against your own goal line
Can you explain why you think this is the case?
Look at the outcomes: Assume you're about 50/50 for making that first down. If you don't, you've given the ball to the other team at 1st and goal.
On the other hand, there's about a 1 percent chance of the other team blocking that punt, a similar chance of it being returned for a touchdown, and a slightly higher chance of the returner muffing the punt and you getting it back.
So, conservatively, let's say you've got a 90 percent chance of the other team getting the ball around midfield vs. a 50 percent chance of the other team getting the ball inside your 10. A lot of it just comes down to whether you like punts, but even I, a known advocate of getting rid of punters entirely, would say that letting your kicker take a whack at it wouldn't be a bad idea.
posted by Etrigan at 8:32 AM on December 9, 2013
Can you explain why you think this is the case?
Look at the outcomes: Assume you're about 50/50 for making that first down. If you don't, you've given the ball to the other team at 1st and goal.
On the other hand, there's about a 1 percent chance of the other team blocking that punt, a similar chance of it being returned for a touchdown, and a slightly higher chance of the returner muffing the punt and you getting it back.
So, conservatively, let's say you've got a 90 percent chance of the other team getting the ball around midfield vs. a 50 percent chance of the other team getting the ball inside your 10. A lot of it just comes down to whether you like punts, but even I, a known advocate of getting rid of punters entirely, would say that letting your kicker take a whack at it wouldn't be a bad idea.
posted by Etrigan at 8:32 AM on December 9, 2013
From the opposition 49 to opposition 30, go for it on fourth-and-4 or less.
From the opposition 29 to opposition 3, go for it on fourth-and-3 or less.
IMO, except in certain situations, when on your opponent's side of the field you should go for it on at any distance that allows you to run a significant portion of your Playbook, at least up to 4th and 10, probably up to 4th and 15.
posted by nathancaswell at 8:33 AM on December 9, 2013
From the opposition 29 to opposition 3, go for it on fourth-and-3 or less.
IMO, except in certain situations, when on your opponent's side of the field you should go for it on at any distance that allows you to run a significant portion of your Playbook, at least up to 4th and 10, probably up to 4th and 15.
posted by nathancaswell at 8:33 AM on December 9, 2013
grog: “This is really cool, but the model is clearly broken for 4th downs when backed up against your own goal line.”I think the reason the bot says go for it is, as is outlined in the piece about the coach that always goes for it on 4th down referenced above, if you punt from 4th and 7 on the 5, the opponent will on average get the ball around 40 yard-line and have a ~77% chance of scoring a touchdown. If you turn it over on downs, they'd have a ~92% chance. If you make it, you keep the ball and obviously they have no chance.
v.q.
Punting Less Can Be Rewarding, but Coaches Aren’t Risking Jobs on It, Adam Himmelsbach, New York Times, 18 August 2012
The 4th Down Study - Part 1
Part 2
Part 3
Part 4, Brian Burke, Advanced NFL Stats, 15-16 September 2009
posted by ob1quixote at 8:39 AM on December 9, 2013 [1 favorite]
Is it measurably easier to break a field-goal record at Denver's altitude than at other NFL cities?
posted by box at 8:39 AM on December 9, 2013
posted by box at 8:39 AM on December 9, 2013
Assume you're about 50/50 for making that first down.
OK, now assume you are 70/30 to make the 1st down because you only need 1 yard. You can go for it and get an expected return of (say).61 points. Kicking has an EP value of -4 points.
Check out this analysis of a 4th and 2 from the 20 early in a game.
posted by Potomac Avenue at 8:41 AM on December 9, 2013
OK, now assume you are 70/30 to make the 1st down because you only need 1 yard. You can go for it and get an expected return of (say).61 points. Kicking has an EP value of -4 points.
Check out this analysis of a 4th and 2 from the 20 early in a game.
posted by Potomac Avenue at 8:41 AM on December 9, 2013
Is it measurably easier to break a field-goal record at Denver's altitude than at other NFL cities?
I know Football Outsiders has 4 difficulty adjustments for their field goal stats, cold, warm, dome and Denver. From their website.
Kickoffs, punts, and field goals are then adjusted based on weather and altitude. It will surprise no one to learn that it is easier to kick the ball in Denver or a dome than it is to kick the ball in Buffalo in December. Because we do not yet have enough data to tailor our adjustments specifically to each stadium, each one is assigned to one of four categories: Cold, Warm, Dome, and Denver.
posted by nathancaswell at 8:43 AM on December 9, 2013
I know Football Outsiders has 4 difficulty adjustments for their field goal stats, cold, warm, dome and Denver. From their website.
Kickoffs, punts, and field goals are then adjusted based on weather and altitude. It will surprise no one to learn that it is easier to kick the ball in Denver or a dome than it is to kick the ball in Buffalo in December. Because we do not yet have enough data to tailor our adjustments specifically to each stadium, each one is assigned to one of four categories: Cold, Warm, Dome, and Denver.
posted by nathancaswell at 8:43 AM on December 9, 2013
Aaaaand Rob Gronkowski's career is basically over. ACL tear with MCL damage to a guy who required 4(!) surgeries for a broken forearm last year.
He required 4 surgeries because of infection, not because of him being injury prone. I think, outside of certain outliers like Bob Sanders or guys who consistently injure the same ankle over and over again, "injury prone" is basically a myth. Any human being who was struck in the knee the way Gronk was struck in the knee would have ligament tears. It's a matter of physics.
posted by nathancaswell at 8:46 AM on December 9, 2013
He required 4 surgeries because of infection, not because of him being injury prone. I think, outside of certain outliers like Bob Sanders or guys who consistently injure the same ankle over and over again, "injury prone" is basically a myth. Any human being who was struck in the knee the way Gronk was struck in the knee would have ligament tears. It's a matter of physics.
posted by nathancaswell at 8:46 AM on December 9, 2013
Three of the 13 FGs over 60 yards were in Denver (Prater's 64-yarder and two 63-yarders). It seems to favor long-but-not-crazy-long FGs, but that may be as much because there are very few attempts at 60 yards, so the data are more anecdotal than useful.
posted by Etrigan at 8:47 AM on December 9, 2013
posted by Etrigan at 8:47 AM on December 9, 2013
Assume you're about 50/50 for making that first down.The model does not seem to make that assumption. It looks like it thinks that going for it on 4th and 1 succeeds 74% of the time and going for it on 2th and 2 succeeds 60% of the time. I assume that this is based on historical data.
posted by dfan at 8:56 AM on December 9, 2013
Sorry, I didn't include Ghost Mode's note from above: "The robocoach would go for it on 4th and 3 FROM ITS OWN 8 YARD LINE." That's the number I was using for the 50/50 chance and to agree that the model is at least a little damaged, if not broken, in that situation.
posted by Etrigan at 8:58 AM on December 9, 2013
posted by Etrigan at 8:58 AM on December 9, 2013
Again, that's just not the case if you look at the numbers (another example). Either way the result is negative points, both going for it and punting. On an average 4th down, going for it yields a less terrible outcome. You can argue with the specifics of a given situation... (They have a great running defense! We have a fantastic defense!) but that just shifts the %s of negative and positive points. It doesn't mean the model is damaged.
posted by Potomac Avenue at 9:03 AM on December 9, 2013
posted by Potomac Avenue at 9:03 AM on December 9, 2013
From that example: Going for it on 4th and 1 succeeds about 74% of the time.
From that position on the field it's worth: -0.2 points (ugh)
But punting yields, on average: -1.0 points
So why wouldn't you go for it?
posted by Potomac Avenue at 9:08 AM on December 9, 2013
From that position on the field it's worth: -0.2 points (ugh)
But punting yields, on average: -1.0 points
So why wouldn't you go for it?
posted by Potomac Avenue at 9:08 AM on December 9, 2013
Using the 4th down calculator from advancednflstats.com to look at 4th and 3 on your own 8:
If you punt, you will be on average down 2.2 points (on average your opponent will score 2.2 points on the following drive starting from about midfield).
If you go for it and fail (43% chance), you will be on average down 4.9 points (with 1st and goal, they'll probably score a touchdown). If you go for it and succeed (57% chance), you'll be on average down 0.2 points (the other team is still a little more likely to score next than you). Weighing those together, you'll on average also be down 2.2 points.
So it's kind of a toss up from an expected value standpoint. Even with the calculator I probably wouldn't go for it myself.
The calculation is unintuitive because of the way we're trained to look at 4th downs, but quantitatively it works out. It's hard to think of turning it over on downs to be "only" as bad as punting followed by the other team driving 40 yards (which happens all the time). Giving up possession of the football automatically turns out to often be quantitatively worse than sometimes giving up possession of the football plus 40 yards.
posted by dfan at 9:18 AM on December 9, 2013 [1 favorite]
If you punt, you will be on average down 2.2 points (on average your opponent will score 2.2 points on the following drive starting from about midfield).
If you go for it and fail (43% chance), you will be on average down 4.9 points (with 1st and goal, they'll probably score a touchdown). If you go for it and succeed (57% chance), you'll be on average down 0.2 points (the other team is still a little more likely to score next than you). Weighing those together, you'll on average also be down 2.2 points.
So it's kind of a toss up from an expected value standpoint. Even with the calculator I probably wouldn't go for it myself.
The calculation is unintuitive because of the way we're trained to look at 4th downs, but quantitatively it works out. It's hard to think of turning it over on downs to be "only" as bad as punting followed by the other team driving 40 yards (which happens all the time). Giving up possession of the football automatically turns out to often be quantitatively worse than sometimes giving up possession of the football plus 40 yards.
posted by dfan at 9:18 AM on December 9, 2013 [1 favorite]
So why wouldn't you go for it?
Public perception. You improve your expected value but the local fans and sports writers will call for your head. Any poker player will tell you it's the right move but the coach knows he'll lose his job on a "risky 4th and 3" call.
As much as they'd like to tell you otherwise, after kickoff, a coach's job is not to lose the game. It's up to the players, the prep, and luck to win it.
posted by cmfletcher at 9:22 AM on December 9, 2013 [1 favorite]
Public perception. You improve your expected value but the local fans and sports writers will call for your head. Any poker player will tell you it's the right move but the coach knows he'll lose his job on a "risky 4th and 3" call.
As much as they'd like to tell you otherwise, after kickoff, a coach's job is not to lose the game. It's up to the players, the prep, and luck to win it.
posted by cmfletcher at 9:22 AM on December 9, 2013 [1 favorite]
So why wouldn't you go for it?That is totally a factor in real life, but here Potomac Avenue was just addressing the question of whether the mathematical model is broken.
Public perception.
posted by dfan at 9:34 AM on December 9, 2013
Again, that's just not the case if you look at the numbers (another example).
Your examples are 4th and 1 from the 29 and 4th and 2 from the 20, and in the latter, the expected value is exactly the same (-1.4 points) if you go for it or if you punt. Push that back 12 yards and add another yard to what you have to get. Is the expected value going to increase in that situation?
posted by Etrigan at 9:37 AM on December 9, 2013
Your examples are 4th and 1 from the 29 and 4th and 2 from the 20, and in the latter, the expected value is exactly the same (-1.4 points) if you go for it or if you punt. Push that back 12 yards and add another yard to what you have to get. Is the expected value going to increase in that situation?
posted by Etrigan at 9:37 AM on December 9, 2013
Public perception. You improve your expected value but the local fans and sports writers will call for your head. Any poker player will tell you it's the right move but the coach knows he'll lose his job on a "risky 4th and 3" call.
That's true, but there's another notable (albeit probably less important) factor, namely, offensive efficiency has gradually drifted up--average yards per play this year is 5.4, up from 4.9 in 1993 and 4.6 in 1973, and turnovers have likewise become less common. This makes going for it more attractive on two dimensions. Most obviously, you are more likely to convert, but also very importantly, it makes control of the ball more valuable and field position less valuable--if you punt and give a team that's highly likely to score the ball at midfield, punting stinks. I suspect that a lot of coaches have a crude mathematical model in their head that's based on what happened in their own playing days that's very hard to overturn. This is pretty common in sports in general--for example, basketball teams have systematically shot far fewer threes than would be suggested as mathematically optimal. It's just very hard to adapt to rule and strategy changes that overturn the way you learn a sport should be played.
posted by dsfan at 9:47 AM on December 9, 2013 [2 favorites]
That's true, but there's another notable (albeit probably less important) factor, namely, offensive efficiency has gradually drifted up--average yards per play this year is 5.4, up from 4.9 in 1993 and 4.6 in 1973, and turnovers have likewise become less common. This makes going for it more attractive on two dimensions. Most obviously, you are more likely to convert, but also very importantly, it makes control of the ball more valuable and field position less valuable--if you punt and give a team that's highly likely to score the ball at midfield, punting stinks. I suspect that a lot of coaches have a crude mathematical model in their head that's based on what happened in their own playing days that's very hard to overturn. This is pretty common in sports in general--for example, basketball teams have systematically shot far fewer threes than would be suggested as mathematically optimal. It's just very hard to adapt to rule and strategy changes that overturn the way you learn a sport should be played.
posted by dsfan at 9:47 AM on December 9, 2013 [2 favorites]
While I think additional reliance on this data is a good thing, no coach wants to be told what to do by a bunch of numbers. I think I've heard this idea somewhere in the past (maybe from Bill Simmons or something), but coaches should hire hardcore Madden players to help them make these decisions. Just fill one of those Gatorade containers with Mountain Dew Baja Blast and they'll probably just show up anyway. Hardcore Madden players have thousands of games under their belt, and have probably dealt with these scenarios more often (albeit in a video game). Plus they could probably help in other areas like clock management, an area I'd guess about a third of NFL coaches are somewhat baffled by.
posted by antonymous at 9:55 AM on December 9, 2013 [1 favorite]
posted by antonymous at 9:55 AM on December 9, 2013 [1 favorite]
Is the expected value going to increase in that situation?
Actually yes, slightly, because the other team has an even better chance of scoring if you punt from the 8 than from the 20!
posted by Potomac Avenue at 9:56 AM on December 9, 2013
Actually yes, slightly, because the other team has an even better chance of scoring if you punt from the 8 than from the 20!
posted by Potomac Avenue at 9:56 AM on December 9, 2013
They'll also have an even better chance of scoring if you fail to make it from the 8 than from the 20!
posted by Etrigan at 9:58 AM on December 9, 2013
posted by Etrigan at 9:58 AM on December 9, 2013
coaches should hire hardcore Madden players to help them make these decisions.
This model has been attempted previously.
posted by cashman at 10:00 AM on December 9, 2013
This model has been attempted previously.
posted by cashman at 10:00 AM on December 9, 2013
Anyway, my point is not to say that the the bot is horrible, or even that coaches shouldn't go for it far more on a wide variety of 4th down situations than they do. My point is that the bot seems, in this particular situation, to be a little bit too "THIS IS WHAT THE INVIOLATE INARGUABLE NUMBERS SAY."
posted by Etrigan at 10:01 AM on December 9, 2013 [1 favorite]
posted by Etrigan at 10:01 AM on December 9, 2013 [1 favorite]
The bot definitely makes a dramatic point! But the key to using it in real world situations (I say this as a poker player heh) is to take the numbers as an average and adjust the inputs to generate strat. So, again in a 4th and 1 on your 8, are you actually lower than the average in terms of likelyhood of making it? Maybe your run offence isn't as good as average, so adjust that % down. But maybe your D is much better than average, so the chances of them scoring from the 50 if you punt is much lower than average too. Make that adjustment in the #s and it nudges the EP lower, making punting a better option. But you can't do that without a statistical average!
posted by Potomac Avenue at 10:17 AM on December 9, 2013
posted by Potomac Avenue at 10:17 AM on December 9, 2013
Offensive numbers have definitely gone up. I just wasted some time plugging average yards per game into excel over the past 10 years and the league wide average per game has gone from around 319 in 2003 ramping up to 348 last year and a pace for over 350 this year.
I suppose the expression should be changed to 3.3 yards and a cloud of fieldturf polymer.
posted by cmfletcher at 10:30 AM on December 9, 2013 [3 favorites]
I suppose the expression should be changed to 3.3 yards and a cloud of fieldturf polymer.
posted by cmfletcher at 10:30 AM on December 9, 2013 [3 favorites]
Yesterday was an especially nutty day in the NFL. ... LeSean McCoy broke the Eagles single-game rushing record in several inches of snow
The crazy stat that I have not seen mentioned (yet) about the Eagles/Detroit game is the absurd ratio of touchdowns to extra-point kicks. There were eight TDs and only a single point-after kick attempt (missed), the other 7 were attempted 2-point conversions.
Has there ever been an NFL game with 0 extra points made for 8 TDs? Or even 8 TDs with a single attempt?
posted by Walleye at 10:37 AM on December 9, 2013 [1 favorite]
The crazy stat that I have not seen mentioned (yet) about the Eagles/Detroit game is the absurd ratio of touchdowns to extra-point kicks. There were eight TDs and only a single point-after kick attempt (missed), the other 7 were attempted 2-point conversions.
Has there ever been an NFL game with 0 extra points made for 8 TDs? Or even 8 TDs with a single attempt?
posted by Walleye at 10:37 AM on December 9, 2013 [1 favorite]
Yesterday was an especially nutty day in the NFL
If we're listing weird things from yesterday, Greg Hardy deserves a mention.
posted by troika at 10:43 AM on December 9, 2013 [6 favorites]
If we're listing weird things from yesterday, Greg Hardy deserves a mention.
posted by troika at 10:43 AM on December 9, 2013 [6 favorites]
If we're listing weird things from yesterday, Greg Hardy deserves a mention.
In keeping with the theme of "These damn kids with their Twitters and bots need to get off my lawn," I will remind you of Alex Karras citing Otis Sistrunk's alma mater as the University of Mars.
posted by Etrigan at 10:57 AM on December 9, 2013 [1 favorite]
In keeping with the theme of "These damn kids with their Twitters and bots need to get off my lawn," I will remind you of Alex Karras citing Otis Sistrunk's alma mater as the University of Mars.
posted by Etrigan at 10:57 AM on December 9, 2013 [1 favorite]
Has there ever been an NFL game with 0 extra points made for 8 TDs? Or even 8 TDs with a single attempt?
There have been only seven games with zero points from kicks. The last one was in 1957.
Total points scored in those five games: 22 (three were 6-0, two were 2-0, two were 0-0).
Total points scored in Eagles-Lions: 54.
posted by Etrigan at 11:31 AM on December 9, 2013 [4 favorites]
There have been only seven games with zero points from kicks. The last one was in 1957.
Total points scored in those five games: 22 (three were 6-0, two were 2-0, two were 0-0).
Total points scored in Eagles-Lions: 54.
posted by Etrigan at 11:31 AM on December 9, 2013 [4 favorites]
The snow helped. (In the slippery, the ball carrier can make the cut easier than the defender can react to it.)There is certainly some truth to this, but for the first half of the game, it was clearly a massive hinderance. When the snow was really coming down hard in the first quarter, McCoy had a grand total of 6 yards rushing. I think he had something like 45 yards at the half, almost all of them on the last drive of the half when the snow was dramatically slowing. He only got another 18 yards in the third, before going wild in the fourth quarter.
posted by Lame_username at 2:13 PM on December 9, 2013 [1 favorite]
It's not that Gronk is fragile, it's that he's likely to be caught in Vegas trying to mosh while still on crutches. Smart and protective of his rehab, this guy is not.
posted by TwoStride at 4:42 PM on December 9, 2013
posted by TwoStride at 4:42 PM on December 9, 2013
There was a guy on talk radio here who reported that Tom Dempsey (1st @ 63 yards) has dementia but when informed he no longer had a tie for the record said "good for him." Google does not have a cite for that but I found this in the New York Times from the New Orleans super bowl coverage on 27 Jan 2013.
posted by bukvich at 6:38 PM on December 9, 2013
posted by bukvich at 6:38 PM on December 9, 2013
Here are some awesome photos that Kyle Grantham took of the snowy Eagles-Lions game and the story of how he shot without autofocus.
posted by meadowlark lime at 11:50 AM on December 10, 2013 [2 favorites]
posted by meadowlark lime at 11:50 AM on December 10, 2013 [2 favorites]
graymouser: "This discussion really isn't complete without Gregg Easterbrook's punting guidelines:"
All football discussions are complete without discussing Gregg Easterbrook.
posted by Chrysostom at 1:45 PM on December 10, 2013 [1 favorite]
All football discussions are complete without discussing Gregg Easterbrook.
posted by Chrysostom at 1:45 PM on December 10, 2013 [1 favorite]
two were 2-0
Oh man, a game where the only points scored were a safety would be AWESOME. Especially if it happened in the 4th quarter.
Safeties are my favorite play. I get all giddy when they happen, even if they're against a team I'm rooting for / have money on.
posted by nathancaswell at 3:46 PM on December 11, 2013
Oh man, a game where the only points scored were a safety would be AWESOME. Especially if it happened in the 4th quarter.
Safeties are my favorite play. I get all giddy when they happen, even if they're against a team I'm rooting for / have money on.
posted by nathancaswell at 3:46 PM on December 11, 2013
The only way safeties would be more awesome is if the team that gave it up were immediately forced to improvise a Dance of Shame. To, of course, Men Without Hats.
posted by Etrigan at 4:50 PM on December 11, 2013
posted by Etrigan at 4:50 PM on December 11, 2013
"Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field" or some such nonsense.
I don't think it's such nonsense. The statistics about going for it on 4th down were created in a context where coaches generally punt on 4th. If the context were different -- if the default was to go for it -- the reality of the game would be different and so too would be calculations. Defenses are playing under the idea that they get the ball back if they stop the offense on 3rd down. Likewise offenses are trying to get 10 yards in 3 downs, knowing that most likely they will kick on fourth if they haven't made to the yard marker. If, suddenly, everybody is thinking it's four downs to get 10 yards, the way each down is played will be significantly different. How exactly is pretty unknowable, but I've no doubt the numbers spit out by the bot would be affected.
posted by 0 at 9:36 AM on December 16, 2013
I don't think it's such nonsense. The statistics about going for it on 4th down were created in a context where coaches generally punt on 4th. If the context were different -- if the default was to go for it -- the reality of the game would be different and so too would be calculations. Defenses are playing under the idea that they get the ball back if they stop the offense on 3rd down. Likewise offenses are trying to get 10 yards in 3 downs, knowing that most likely they will kick on fourth if they haven't made to the yard marker. If, suddenly, everybody is thinking it's four downs to get 10 yards, the way each down is played will be significantly different. How exactly is pretty unknowable, but I've no doubt the numbers spit out by the bot would be affected.
posted by 0 at 9:36 AM on December 16, 2013
"Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field" or some such nonsense.
I don't think it's such nonsense. The statistics about going for it on 4th down were created in a context where coaches generally punt on 4th. If the context were different -- if the default was to go for it -- the reality of the game would be different and so too would be calculations.
I agree 100 percent, but the default stance of 99 percent of people employed in any sport where a new statistic seems to overturn the conventional wisdom is, "Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field."
posted by Etrigan at 9:49 AM on December 16, 2013
I don't think it's such nonsense. The statistics about going for it on 4th down were created in a context where coaches generally punt on 4th. If the context were different -- if the default was to go for it -- the reality of the game would be different and so too would be calculations.
I agree 100 percent, but the default stance of 99 percent of people employed in any sport where a new statistic seems to overturn the conventional wisdom is, "Yeah, that's just statistics. It doesn't capture the flow of the game and the reality down on the field."
posted by Etrigan at 9:49 AM on December 16, 2013
I didn't mean to pick at your comment specifically; just something I thought about yesterday while watching games with this thread in the back of my head and your quote was a good jumping off point. I didn't have answer to the question of "if the stats say going for it is advantageous, then why don't more coaches do it?" The common answer here that it's because coaches are conservative didn't really feel right to me, but I couldn't put my finger on why or propose an alternative. This weekend the alternative of "the stats reflect the context and may imply different conclusions if the context were different" finally occurred to me.
posted by 0 at 10:02 AM on December 16, 2013
posted by 0 at 10:02 AM on December 16, 2013
Now that I think of it, I think the stats probably wouldn't change much if teams went for it on fourth down more frequently. It's not like the defense doesn't know they're going to go for it when it happens (at least, in my memory, fake punts/FGs are considerably rarer than straight-up "They're going to go for it"), and they have just as much time to prepare as the offense does. The success numbers would drop, but not terribly much.
posted by Etrigan at 11:20 AM on December 16, 2013
posted by Etrigan at 11:20 AM on December 16, 2013
« Older What Young Gay Men DO Know About AIDS | The Voice of Russia Newer »
This thread has been archived and is closed to new comments
posted by snarfles at 6:32 AM on December 9, 2013