Pete Carroll is getting a lot of grief today about going for it on 4th and 1 from the ATL 11 in the 2nd qtr. It was the slam-dunk right call, which I'll explain below. But worse, much of the criticism is completely misguided. I would, however, knock SEA for the play call.
I keep reading and hearing about how the failed 4th down attempt 'came back to bite Seattle' or that the decision cost 3 points in a game decided by 2. First of all, that's complete outcome bias. Expunge that kind of thinking from your head. Had SEA converted, the same people who are now critical would be praising Carroll for his courage. Second, you can't simply append 3 extra points onto the final score and say that would have changed the outcome. You never know how the game may have unfolded had Carroll decided on the FG attempt. Both teams would have played at different risk levels and at different paces had the score been different, particularly in the endgame.
I've refrained from the cookie-cutter 'should have gone for it, coach' posts this season for fear of becoming a one-trick pony. (Especially now that everyone can do the same thing himself with the 4th Down Calculator.) But this was a critical call in an important game, so I'll indulge. Here are the numbers on the 4th down itself. Remember, results are based on league average baseline numbers. It was 4th and 1 on the ATL 11 with 5:38 in the 2nd qtr. ATL was up 13-0.
Based on point expectancy:
A successful FG is worth 2.4 EP.
A missed FG is worth -0.26 EP.
FGs are good from that range 93% of the time, making the total expected value of the FG attempt:
.93 * 2.4 + (1-.93) * -.26 = 2.20 EP
A successful conversion is worth 4.74 EP.
A failed conversion is worth 0.16 EP.
Conversions are successful 74% of the time, making the total value of the conversion attempt:
.74 * 4.74 + (1-.74) * 0.16 = 3.55 EP
The numbers are strongly in favor of going for it. In fact, you'd need only a 45% success rate to justify the conversion attempt. Further, SEA was down by 13 at the time, and needed some breaks to go their way to get back into the game. Let's see what the win probability model says:
Based on win probability:
A successful FG is worth 0.16 WP.
A missed FG is worth 0.11 WP.
FGs are good from that range 93% of the time, making the total expected value of the FG attempt:
.93 * .16 + (1-.93) * .11 = 0.16 WP
A successful conversion is worth 0.25 WP.
A failed conversion is worth 0.11 WP.
Conversions are successful 74% of the time, making the total value of the conversion attempt:
.74 * .25 + (1-.74) * .11 = 0.21 WP
Again, the numbers favor going for it. Using WP shows that the break-even conversion probability is 33%, even lower than for the EP analysis.
The play call:
I continue to be baffled at why teams line up in jumbo goal line personnel and formation for 4th and short plays only to run a simple RB or FB dive. Spread it out. Force the defense to defend the entire field. Take advantage of the fact the defense has 10 guys crowding the line of scrimmage. The dive play is completely predictable and plays right into the hands of the defense. Remember that a 4th and short from the middle of the field is very different than on the goal line. There is no value to the yards beyond the goal line, but there is value beyond the line to gain. The jumbo dive play is probably the right call on the goal line, but not 11 yards away from it.
The Seahawks' 4th and 1
By
Brian Burke
published on 1/14/2013
in
4th down,
analysis,
game analysis,
strategy
Subscribe to:
Post Comments (Atom)
Yeah, I'm with you. This one was so obvious that I glossed over it in my weekend recap because well, what's the point? If you're arguing that it was the wrong call, my view is I don't have the desire to argue with you.
I'll just add that Seattle had signed a kicker off the street, which makes the decision even more obvious. Not significant, but maybe instead of 93% the field goal is only good 87% of the time or something.
How about Atlanta's failure to go for two when up by 19 after Seattle had two offsides calls on the extra point attempt?
Never understand why the give it to the upback, either. Rarely works.
I absolutely love the outcome bias reference. That is one that I will use with my buddies for the rest of my life. I'll get back to you on the success rate of adoption of the bias among them...
How do you get 2.4 EP for a successful FG? I read the article and comments here:
http://www.advancednflstats.com/2010/01/expected-points-ep-and-expected-points.html
I guess it's because the following kickoff is worth -0.6 EP? But according to the EP article, EP takes into account *all* future drives by both teams. I can see taking the very next drive's EP, but how can you account for all future drives by both teams, wouldn't that be unknown/infinite? 3.0 - 0.6 + ? + ? + ...
Teams sometimes give it to the fullback (or upback) to 1) break tendencies and 2) not allowing the opposing DL and LBs as much time to penetrate and stop for a loss or no gain. Sometimes it's best to give it to the FB who is one yard to two yards deep rather than the RB who is 4 yards deep and has to travel 5 yards total to get a first down.
I'm kinda surprised they didn't try a Wilson QB sneak on one of the short yardage situations...he could come up to the line and then pat the center on his right butt or left butt to signal which way he wants to go. Even QB sneaks behind the guard or even Okung often work because the DL and LBs are pinching inwards to stop the dive.
Some more basic numbers that support what you're saying abotu it being the right call -- Seattle made the 5th most 4th down attempts this season (18), and was the 9th most successful in terms of percentages (61%). Sure that's small sample size, but, it shows they've been pulling it off consistently. They had every reason to believe they could convert, and in that case, they should go for it.
imo-I'd recommend reading the explanation of why we account for the value of the subsequent kickoff in the Romer 4th down study.
I would not try a QB sneak either. You've got over 1800 lbs of humanity submarining into your o-line to prevent that. With a running QB, I'd do a shotgun play fake, pass to a slant or quick, in as my base short yardage set.
It's worth pointing out that the extra yards after getting a first down on the opponent's 11 yard line are also some of the most valuable on the field because they make for a progressively easier first and goal.
My data doesn't tell me how long that last yard is. Or, for that matter, how favorable the spot will be. I think those are worth taking into considerations when making go/no go decisions. The jumbo package probably works great when all you need is an inch.
4th and 1 and the field goal both seem to be less successful that what you have, I see a conversion rate of 1476 of 2247 -- around 66% - since 2001. I also see a lower field goal rate -- around 89%.
I've got to ask: why is the value of a failed 4th down attempt (0.16) so much higher than the value of a missed field goal (-0.26)?
That said, going for it is not a marginal choice in this scenario, fudging the numbers a bit won't change that, and Pete Caroll should be lauded for making the correct decision.
Nate-4th downs are often attempted by inferior teams in desperation mode. There are some ways to correct for that. Pls see the series on 4th downs.
Remember that missed FG attempts inside the 20 move the ball out to the 20. The EP value difference is the difference between giving the ball to your opponent on the 11 vs on the 20.
Of course, none of that accounts for getting the goose egg off the scoreboard.
Can someone put the 4th down calculation into context for me? From the article, we know that "results are based on league average baseline numbers." But this wasn't the average NFL offense going against the average NFL defense. It was a specific offense (Seattle) going against a specific defense (Atlanta). It seems to me that the quality of your offense/the opponent's defense is going to greatly affect the odds of successfully converting on a given play.
For example, the odds of my team successfully converting the 4th down attempt against the Steelers is not the same as the odds of converting against the Saints. Similarly, all else being equal, the Redskins have a better chance of converting than do the Cardinals.
As far as I can tell, the calculator doesn't take these factors into consideration. The calculation is based upon league averages. If that's correct, doesn't it call into question the value of the calculator?
I am sure that I am missing something obvious here, so please excuse my ignorance and be gentle.
If Atlanta had gone for two successfully when up 26-7, the score would have been 28-28 when they got the ball on their own 28 with 19 seconds left and they probably would have taken a knee and gone to overtime. (or is that outcome bias)
> ..., doesn't it call into question the
> value of the calculator?
Really, that depends a lot on what you think the value of the calculator is. Sure, the calculator is a 'best guess', but it's the best one we have. (Feel free to show us a better one. Michael Beuoy is trying to compensate by using point spread values. I did a regression w.r.t. third and 1 conversions recently, but team-by-team data is too noisy for specifics like this.)
A lot of the value of the calculator is having those rough values. In this scenario, it's not close: Going for it is the right call even if the field goal were a sure thing, and the team could only covert the fourth and 1 half of the time.
I'm on a phone here so can't check the calculator, but doesn't the calculator give the baseline for the success rate needed for the attempt to be a smart call. In this case the break even point was 33%. When coach has this information he can then estimate his teams chances of converting. If the probability is higher: go for it. If the probability is lower: kick it.
Pete Carroll actually made it to #1 of the shame report for his "terrible" decision to go for it.
http://www.nfl.com/videos/the-shame-report/0ap2000000127102/The-Shame-Report-Divisional-Weekend
My reaction at the time was that it was absolutely the right decision, even after they failed to convert.
My only objection was the playcall, and that I thought they could have played the series differently (assuming they were planning to go for 4th and 1 the whole time if they failed on third down). It seemed to me that a shot to the endzone on 3rd down or something other than a run up the middle would have had a higher-than-normal probability for success, given the down and distance.
This just in: win probability is an estimate! A freaking estimate! Fact is, they had a 100% chance of not converting, as well as a 100% chance of losing the game. As long as we're basing analysis of a game on prior results, let's go all in with the approach.
Anonymous - Here is the post NateTG referenced: 4th and 1 Conversion
As best as I can tell, 4th and 1 and 3rd and 1 conversions are a weighted coin flip, where the weight doesn't seem to vary much from team to team. Good teams are slightly better at it than bad teams, but only on the order of +/-5%.
But on one single 4th down, percentages don't matter; how many yards you gain is all that matters. Being a successful NFL coach is not purely a science.. It is an art. Which is why you don't see many stats geeks win Super Bowls as head coaches.
Funny how immediately after citing outcome bias as a problem and systematic error in decision making someone comes along, frothing at the mouth, to illustrate this point.
The comment about 3rd and 1 was more just an intuition about a playcalling against a defense that sells out to stop 3rd and short (under the assumption that the defense believes the offense will kick the FG on 4th down), and exploiting that.
The seahawks have run this identical play numerous times this year. If you see short yardage, I-form, esp 4th down, this was THE play to bet on. I read somewhere where someone speculated that this one particular play was likely thr #1 repped play by the falcons in this situation. They were ready, hawks should have broken tendency.
Someone talked about the play sequence given assumption that they knew they would use 4 downs. I think problem is they didnt know. Carroll is relying on gut impulse rather folliwing a clear strategy.