GDT: Super Bowl 49

[bluedevil58*]

 

[Ole Gil]

I understand the opposing argument. I simply think it was the wrong decision. The coach played that 2nd down under the assumption that he would have two more plays to run after it. That's a foolish assumption, especially with the title on the line.

This is false. He did it under the assumption he'd have a 98-99.5%ish chance of 2 more plays to run after it, which was correct.

 

[Blueline Bomber]

Since there's fancy stats for the "Pass was correct" side...

http://www.cbssports.com/nfl/eye-on-football/25017292/super-bowl-49-how-unlikely-was-russell-wilsons-interception

The Seahawks were a phenomenally successful power-running team this season. According to Football Outsiders, in fact, they were the second-best power team in the league, converting 81 percent of power runs (described as third or fourth down and/or any goal-to-go situation with 2 yards or less to go) into first downs or touchdowns.

In the regular season and playoffs combined, Seattle had run 89 plays with 2 yards or less to go (in games separated by two touchdowns or less) before that fateful pass. They had called for a run on 66 of those plays (74.2 percent), compared to only 23 pass attempts. If we widen the sample to the three years that Russell Wilson has been in the league, they had called 181 runs to 75 passes in similar situations, a run rate of 70.7 percent.

They had faced five similar situations earlier in the Super Bowl alone and had given the ball to Lynch four of those times, with him converting once for a touchdown and once for a first down (Note: The only throw was an incomplete pass to Jermaine Kearse, who was being covered by ... Malcolm Butler). This is all by way of saying that this specific Seahawks team even calling a pass play in this situation was an extremely rare occurrence.

Since 1998, teams had been faced with second-and-goal from inside the 2-yard line, with less than five minutes left in a game where they were down between four and eight points (so it is still a one-possession game but they are in need of a touchown) 81 times. Of those 81 plays, 47 were runs and 34 were passes. That's a pass play rate of 42 percent, which is not all that low, but it's significantly higher than the rate at which Seattle called passes in 2-yards-to-go situations this season and over the past three.

On those 47 running plays, teams had scored 25 touchdowns and fumbled twice. That's a touchdown rate of 53.2 percent. Pretty good stuff. Considering it was Marshawn Lynch in the backfield, I think it's fairly safe to say Seattle's odds of scoring if they had run the ball might have been even higher than that.

Meanwhile, on those 34 pass plays, teams had gone 14 of 33 for 14 touchdowns and one sack. That's a touchdown rate of 41.2 percent, significantly lower when compared to the rate of teams that ran the ball in similar situations.

In other words, even statistically speaking, he would have been better off running the ball.

 

[Ole Gil]

Since there's fancy stats for the "Pass was correct" side...

http://www.cbssports.com/nfl/eye-on-football/25017292/super-bowl-49-how-unlikely-was-russell-wilsons-interception

In other words, even statistically speaking, he would have been better off running the ball.

A lot of people agree with you, so there's that, but you're missing all kinds of stuff. If you don't know what to do with the numbers, they don't do you any good.

 

[Anton Dubinchuk]

Since there's fancy stats for the "Pass was correct" side...

http://www.cbssports.com/nfl/eye-on-football/25017292/super-bowl-49-how-unlikely-was-russell-wilsons-interception





In other words, even statistically speaking, he would have been better off running the ball.

either you dont know what those statistics mean or youre missing the point. no one said they were more likely to score on that play by passing. all were saying is that its better to go pass run run than run run.

in fact all that last stat tells us is that it was a higher percentage play, because in those situation no one had EVER thrown a pick, while they had fumbled 2 out of 47 times. if anything, the stats say that the pass was safe there and they still couldve run it two times. so if anything, they strengthen the argument, which the whole time has been that three plays are better than two.

 

[Blueline Bomber]

But it wasn't running 3 plays or 2. Because, again, that assumes you get 3 plays or 2. All that matters is what you should do on THAT play, because THAT play was the only guaranteed one you have.

Yes, if he was guaranteed 3 plays, pass/run/run was correct. But since that was impossible to guarantee, you run the play that is most likely to get you in the endzone and put your team ahead in the final seconds of the Super Bowl. Which was a run.

 

[Anton Dubinchuk]

But it wasn't running 3 plays or 2. Because, again, that assumes you get 3 plays or 2. All that matters is what you should do on THAT play, because THAT play was the only guaranteed one you have.

Yes, if he was guaranteed 3 plays, pass/run/run was correct. But since that was impossible to guarantee, you run the play that is most likely to get you in the endzone and put your team ahead in the final seconds of the Super Bowl. Which was a run.

using your stats, its actually probably a fairly easy probalistic model (granted, an incomplete one) that says you're wrong.

 

[HisIceness]

One thing I think most of us can agree on, at least Joe Buck wasn't calling the game.

 

[Anton Dubinchuk]

ok, super simple probabilistic model coming up.

assumptions that i will make (because obviously there have to be assumptions):

- i will assume that the data above hold not just for 2nd down, but for 3rd and 4th as well (no reason it shouldnt, save for typical small variation)

- for the pass data, it didnt state that there were any picks. considering it stated that there was a sack, im assuming thats because there were no picks. that being said, as a concession to your point, i will consider yesterdays pick as a part of the data, even though thats the very event we are modeling. i also dont know what to do with the sack, so as another concession to your argument i will consider it another pick. that means, 35 pass plays, 14 tds, 2 turnovers (for the sake of the math). 40% td, 5.7% turnover.

- i agree with the writer's assessment that lynch probably wouldve scored more often than the norm. ill give it an even 65%, if that seems fair to you. ill keep the turnover data where it is, so 65% td, 4.25% turnover.

after that it becomes pretty simple. run two possible patterns, pass run run, and run run, and see which one scores more often.


the data for pass run run is as follows:

score on first play: 40%
turnover on first play: 5.7%
score on second play: 35.3%
turnover on second play: 2.3%
score on third play: 10.9%
turnover/dont score on third play: 5.8%

total chance of scoring a touchdown: 86.2%


the data for run run is as follows:

score on first play: 65%
turnover on first play: 4.25%
score on second play: 20%
turnover/dont score on second play: 10.75%

total chance of scoring a touchdown: 85%





ill admit this isnt a very scientific or thorough model, but it gets the point across. youve got a higher chance of scoring, giving the data you provided, by passing the ball on second down. i get that its a simple model, but i even gave you concessions as far as the turnover chances for passing goes, considering last night was literally the first time its happened with those parameters.

you can say that this isnt conclusive, and i agree, anything that i can do with a pen and paper in 10 minutes is probably not too exhaustive. but to say that the stats support your argument because "you dont get guaranteed a second play, just a first" is silly and not how things work. now, an argument saying that they couldve squeezed 3 plays into those 26 seconds is a different argument, different discussion. but as far as being "not guaranteed more plays," its basically just high school probability.

 

[Blueline Bomber]

using your stats, its actually probably a fairly easy probalistic model (granted, an incomplete one) that says you're wrong.

Again, Seattle had an 81% conversion rate when running the ball for a first down/touchdown when dealing with 2 or less yards. They decide to run in those situations 70% of the time, meaning the players are more comfortable running than passing there. The league, as a whole, has converted touchdowns at a higher rate using the run than using the pass in those situations.

All of which serve as evidence as to why the run would have been a better play there. Yes, the coach's reasoning was solid if you work under the assumption that he had 3 plays to work with. Since he didn't, it falls apart.

I wonder how many times a team has taken the lead in the final minute of a game and held onto it to win the game? Since the reasoning he gave for wanting to run the 3 plays is to run the clock to 0.1. Again, the reasoning is solid, since you don't want to give the other team the ball back, but given Seattle's defense, getting the points and defending the lead seems like a more solid plan.

 

[Unsustainable]

 

[Blueline Bomber]

I should also add the reason I focused on only the initial play is because there are a myriad of factors that could have affected the clock/distance in the possible 2nd/3rd plays. A false start, for instance, on the 2nd down would essentially force Seattle to throw for all 3 plays (assuming they passed on the first and it was simply incomplete rather than picked).

 

[tarheelhockey]

Man, I'm FAR from a fancy-stats h8r type of fan.

But when it gets to the point that you're justifying a slant pass on the 1 to win the Super Bowl, you're doing it wrong.

I don't give a **** if the pass gives them 2 more plays. That has zero relevance because they have Lynch and a timeout on the 1 in the ****ing Super Bowl. Throwing in that situation is supposed to be a "safe" play, are you kidding me?

It's not a safe play because you are laying the ball out there for the defense to intercept and steal a championship right from out of your grasp... which is exactly what happened. The coach tried to be a smart guy and play ridiculous percentages three plays ahead so he could kill 10 meaningless seconds, and he lost the ****ing Super Bowl.

Seriously, this "results-based thinking" stuff has its place but it becomes ridiculous after a point. This was a glaringly horrible call that shocked every football fan, player and coach in the world. Bending backwards to prove that it was the "right" call because some obscure stat says so... come on, man. Sometimes sports really is as simple as, "here's the ball, there's the goal line, man up and win the game."

 

[Anton Dubinchuk]

Man, I'm FAR from a fancy-stats h8r type of fan.

But when it gets to the point that you're justifying a slant pass on the 1 to win the Super Bowl, you're doing it wrong.

I don't give a **** if the pass gives them 2 more plays. That has zero relevance because they have Lynch and a timeout on the 1 in the ****ing Super Bowl. Throwing in that situation is supposed to be a "safe" play, are you kidding me?

It's not a safe play because you are laying the ball out there for the defense to intercept and steal a championship right from out of your grasp... which is exactly what happened. The coach tried to be a smart guy and play ridiculous percentages three plays ahead so he could kill 10 meaningless seconds, and he lost the ****ing Super Bowl.

Seriously, this "results-based thinking" stuff has its place but it becomes ridiculous after a point. This was a glaringly horrible call that shocked every football fan, player and coach in the world. Bending backwards to prove that it was the "right" call because some obscure stat says so... come on, man. Sometimes sports really is as simple as, "here's the ball, there's the goal line, man up and win the game."

im not justifying the slant. ive already said that like 10 times. im just justifying a pass play. again, i wouldve floated it to matthews back corner.

 

[Anton Dubinchuk]

All of which serve as evidence as to why the run would have been a better play there. Yes, the coach's reasoning was solid if you work under the assumption that he had 3 plays to work with. Since he didn't, it falls apart.

I wonder how many times a team has taken the lead in the final minute of a game and held onto it to win the game? Since the reasoning he gave for wanting to run the 3 plays is to run the clock to 0.1. Again, the reasoning is solid, since you don't want to give the other team the ball back, but given Seattle's defense, getting the points and defending the lead seems like a more solid plan.

my math was done assuming that a turnover ends the drive. percentages still end up higher.

and obviously that was a part of the plan but not the whole plan, bringing the time down to 1 sec. if they actually wanted to throw the play away, they'd spike the ball. he was trying to give his team their best chance to win by running three plays. and more often than not, it would work out. it didnt, so people are mad. wilson puts that ball in the right place and no ones saying anything.

 

[Surrounded By Ahos]

One thing we can all agree upon, this was ****ing hilarious:

 

[Blueline Bomber]

^ Yes

my math was done assuming that a turnover ends the drive. percentages still end up higher.

All your math shows is that having an additional play gives a better odd of scoring a touchdown. That having 3 shots at the end zone has a better chance of scoring a touchdown than 2. Nothing ground-breaking there.

Your math showed that in the first play of the drive, there was a 65% chance of the run scoring. That, combined with Seattle's 81% conversation rate of short-down power running, is all I needed to see. If that scores, the odds of scoring on the next two plays don't matter.

If you have a 65% chance of scoring on a run, or a 40% chance of scoring on a pass (and a lower turnover rate on the run play as well), why in the world would you attempt to justify the pass as the correct decision?

 

[tarheelhockey]

This is brilliant.

 

[Anton Dubinchuk]

^ Yes



All your math shows is that having an additional play gives a better odd of scoring a touchdown. That having 3 shots at the end zone has a better chance of scoring a touchdown than 2. Nothing ground-breaking there.

Your math showed that in the first play of the drive, there was a 65% chance of the run scoring. That, combined with Seattle's 81% conversation rate of short-down power running, is all I needed to see. If that scores, the odds of scoring on the next two plays don't matter.

If you have a 65% chance of scoring on a run, or a 40% chance of scoring on a pass (and a lower turnover rate on the run play as well), why in the world would you attempt to justify the pass as the correct decision?

because you get three of them. if you dont pass, you get two.

accepting of my math or not, you are revealing to me here that your ability to interpret statistics leaves much to be desired. the last paragraph of your post is not making your case very well here. you are essentially saying "why would i want to flip a fair coin twice to try to get heads when i could flip an unfair coin once which turns up heads 70% of the time."

essentially, you are saying my math isnt groundbreaking (agreed), but then completely ignoring the very part about it that isnt groundbreaking. a thing that isnt groundbreaking doesnt stop being true by the nature of not being groundbreaking.

 

[Blueline Bomber]

because you get three of them. if you dont pass, you get two.

accepting of my math or not, you are revealing to me here that your ability to interpret statistics leaves much to be desired. the last paragraph of your post is not making your case very well here.

All it tells me is that you're treating it as a purely statistical standpoint and not taking into account that we're talking about an actual game here. If you told any coach that he could win the Superbowl with a touchdown and offered him the choice between two plays, one with a 65% chance of success and one with a 40% chance, what play do you think will be picked?

Getting 3 or 2 plays doesn't matter if you score. The point, ultimately, was to score the game-winning touchdown. Your own statistics showed that they had a significantly better chance of scoring if they ran the ball initially. There's not much to interpret there.

 

[Anton Dubinchuk]

All it tells me is that you're treating it as a purely statistical standpoint and not taking into account that we're talking about an actual game here. If you told any coach that he could win the Superbowl with a touchdown and offered him the choice between two plays, one with a 65% chance of success and one with a 40% chance, what play do you think will be picked?

Getting 3 or 2 plays doesn't matter if you score. The point, ultimately, was to score the game-winning touchdown. Your own statistics showed that they had a significantly better chance of scoring if they ran the ball initially. There's not much to interpret there.

if i asked any coach if he had two attempts at 65% or an attempt at 40% and THEN two at 65% he would take the latter every time because he probably studied statistics at above an 8th grade level. my own statistics did NOT show that, again, you are showing me that you dont remotely know how to interpret statistics.

 

[Blueline Bomber]

if i asked any coach if he had two attempts at 65% or an attempt at 40% and THEN two at 65% he would take the latter every time because he probably studied statistics at above an 8th grade level. my own statistics did NOT show that, again, you are showing me that you dont remotely know how to interpret statistics.

Well, like you said earlier, we'll agree to disagree. Planning on running 3 plays to earn a 1.2% better chance of scoring than running 2 doesn't make a hell of a lot of sense. Especially when you consider mitigating factors that won't show up on statistics, like the possibility of a penalty and the subsequent pressure/confidence of Seattle/New England players.

Like I said earlier, if 81% of the time, Seattle had converted on 2 or less yards by running the ball, that's all I would need to hear. Don't worry about the clock, run the ball, get the score, and let your defense win you the game.

 

[Anton Dubinchuk]

Well, like you said earlier, we'll agree to disagree. Planning on running 3 plays to earn a 1.2% better chance of scoring than running 2 doesn't make a hell of a lot of sense. Especially when you consider mitigating factors that won't show up on statistics, like the possibility of a penalty and the subsequent pressure/confidence of Seattle/New England players.

Like I said earlier, if 81% of the time, Seattle had converted on 2 or less yards by running the ball, that's all I would need to hear. Don't worry about the clock, run the ball, get the score, and let your defense win you the game.

fair enough. worth noting that i turned that sack into an interception to help your case. basically, part of those "other factors" where they would lose yards were basically just turned into turnovers to make the numbers closer. too simple a model, but the concessions i made i would say fairly put the "other factors" AGAINST the pass, and it still worked out in their favor. to me, taking an action that, all things considered, gives you an additional 1% chance of scoring makes all the sense in the world. if you ask a coach if he wants an 85% chance of scoring a td, or an 86% chance, he would take the 86%, most of the time.

 

[Blueline Bomber]

I'd think the coaches would want to leave less to chance (hence why he wanted to run the clock down to .1 in the first place). Is 1% really worth running an additional play and risk having something go wrong?

 

[Anton Dubinchuk]

I'd think the coaches would want to leave less to chance (hence why he wanted to run the clock down to .1 in the first place). Is 1% really worth running an additional play and risk having something go wrong?

fair enough. good talk.