4th and 4: Hauschka 53 Yard attempt decision - simple analytics

[IgotBILLStopay]

(click on the decision tree picture below to view a bigger version of the image)

 

McD admitted that they talked about whether to go for it on 4th and 4 or attempt the 53 yarder. Obviously they decided on the latter. I am revisiting that decision in this post since it appears very strange from a basic analytics perspective. That decision should be made based on 6 probability assumptions. 

 

1. The probability that Hauschka makes the 53 yarder (he had missed his previous 3 from 50+ yards and missed a 34 yarder earlier in the game - but holds the record for consecutive makes of 50+ yards)

2. The probability that the Bills convert a 4th and 4

3. The probability that the Bills score a TD in 15 seconds with a TO after converting the 4th down

4. The probability that the Hauschka makes a shorter field goal after the conversion on 4th and 4 

5. The probability that Bills win in OT

6. The Browns with a TimeOut and 22 seconds can get in position for a FG and make it

 

Here are my estimates:

 

1. 30% 2. 46% (https://blog.minitab.com/blog/the-statistics-game/calculating-the-probability-of-converting-on-4th-down)  3. 30% 4. 70% 5. 50% 6. 10%

 

 

That puts the win probability at 13.5 % for the 53 yard try  vs. 25.3% for going for it). The probability of a successful 53 yarder has to be close to 60% for the Bills to be comfortable going for it even keeping other probabilities same (not a given - if one has a 60% confidence in a Hauschka 53 yarder - the probability of a make from a shorter distance will be much higher than the assumed 70% - that will still suggest going for it on 4th down ). Listen - I continue to believe we lucked out in having a great coach who is able to prepare the players well and motivate them to play for each other. That being said, the decision to kick a 53 yarder appears quite short-sighted.

 

Dont the Bills have a basic analytics guy to advise them?

Edited November 11, 2019 by IgotBILLStopay

[ALLEN1QB]

I have no problem with them trying to kick the field goal what I have a problem with is the stupid play on 3rd down why not try to get him a little closer instead they throw a low percentage pass that we know Josh Allen cannot hit. Right now Buffalo should have one loss and that would be the one to the Eagles the other to the coaches lost with stupid play calls throughout the game

Edited November 11, 2019 by ALLEN1QB

[boater]

3 minutes ago, IgotBILLStopay said:

Dont the Bills have a basic analytics guy to advise them?

I think their analytics is done during the week and focuses on bigger questions than forth and goal. I suspect there  is not a game day analytics operation that works on a down by down basis.

 

Nice work on the analysis. I didn't know Minitab was still a thing. I thought everybody moved to R. I used Minitab decades ago in a quality assurance application.

 

 

[The Jokeman]

Just now, ALLEN1QB said:

I have no problem with them trying to kick the field goal what I have a problem with is the stupid play on 3rd down why not try to get him a little closer instead they throw a low percentage pass that we know Josh Allen cannot hit that's what I have a problem with. 

Exactly on 3rd and short we should have tried to take advantage of quickness of guys like Beasley and Singletary. 

[Mark80]

30% chance they score a td with 15 seconds left seems way high to me.

Edited November 11, 2019 by Mark80

[dorquemada]

1 minute ago, boater said:

I think their analytics is done during the week and focuses on bigger questions than forth and goal. I suspect there  is not a game day analytics operation that works on a down by down basis.

 

Nice work on the analysis. I didn't know Minitab was still a thing. I thought everybody moved to R. I used Minitab decades ago in a quality assurance application.

 

 

 

I'm  not sure there's a bigger question than 4th and 4 for the game.

 

Additionally, whatever the Bills' analytics team is working on, they're not doing a very good job.  Unless, of course, Daboll's super cute and clever play calling is a result of analytics.  For example, when facing the #30 run defense, make sure to pass 75% of the time, they'll never know what hit them!

[GG]

1 minute ago, boater said:

I think their analytics is done during the week and focuses on bigger questions than forth and goal. I suspect there  is not a game day analytics operation that works on a down by down basis.

 

Nice work on the analysis. I didn't know Minitab was still a thing. I thought everybody moved to R. I used Minitab decades ago in a quality assurance application.

 

 

If they don't have this kind of data to help in-game decisions, they should.

[Buffalo Barbarian]

13 minutes ago, IgotBILLStopay said:

(click on the decision tree picture below to view a bigger version of the image)

 

McD admitted that they talked about whether to go for it on 4th and 4 or attempt the 53 yarder. Obviously they decided on the latter. I am revisiting that decision in this post since it appears very strange from a basic analytics perspective. That decision should be made based on 6 probability assumptions. 

 

1. The probability that Hauschka makes the 53 yarder (he had missed his previous 3 from 50+ yards and missed a 34 yarder earlier in the game - but holds the record for consecutive makes of 50+ yards)

2. The probability that the Bills convert a 4th and 4

3. The probability that the Bills score a TD in 15 seconds with a TO after converting the 4th down

4. The probability that the Hauschka makes a shorter field goal after the conversion on 4th and 4 

5. The probability that Bills win in OT

6. The Browns with a TimeOut and 22 seconds can get in position for a FG and make it

 

Here are my estimates:

 

1. 30% 2. 46% (https://blog.minitab.com/blog/the-statistics-game/calculating-the-probability-of-converting-on-4th-down)  3. 30% 4. 70% 5. 50% 6. 10%

 

 

That puts the win probability at 13.5 % for the 53 yard try  vs. 25.3% for going for it). The probability of a successful 53 yarder has to be close to 60% for the Bills to be comfortable going for it even keeping other probabilities same (not a given - if one has a 60% confidence in a Hauschka 53 yarder - the probability of a make from a shorter distance will be much higher than the assumed 70% - that will still suggest going for it on 4th down ). Listen - I continue to believe we lucked out in having a great coach who is able to prepare the players well and motivate them to play for each other. That being said, the decision to kick a 53 yarder appears quite short-sighted.

 

Dont the Bills have a basic analytics guy to advise them?

 

they were 0% on 4th down conversions earlier in the game. What do the analytics say about that ??

 

 

[IgotBILLStopay]

2 minutes ago, Mark80 said:

30% chance they score a td with 15 seconds left seems way high to me.

Fair enough. Would you be comfortable with 20% there? The revised win probability assuming 20% is 22.08%. It is still way higher than kicking a 53 yarder.

 

The basic intuition is that kicking the field goal only ties the game - you have to then win it in OT whereas going for it gives you a chance to either win it outright or attempt a higher probability kick.

[H2o]

9 minutes ago, ALLEN1QB said:

I have no problem with them trying to kick the field goal what I have a problem with is the stupid play on 3rd down why not try to get him a little closer instead they throw a low percentage pass that we know Josh Allen cannot hit. Right now Buffalo should have one loss and that would be the one to the Eagles the other to the coaches lost with stupid play calls throughout the game

[IgotBILLStopay]

10 minutes ago, boater said:

Nice work on the analysis. I didn't know Minitab was still a thing. I thought everybody moved to R. I used Minitab decades ago in a quality assurance application.

 

 

 

All I did was google 4th down and distance success percentage. I guess google still goes to the Minitab blog:)

[uticaclub]

We beat the Jets & Titans because they missed FGs. It's just coming full circle in the second half of the season, now it's our time to lose games because of missed FGs.

[IgotBILLStopay]

According to this post (which uses R), the probability of making a 53 yarder with the game on the line is likely even lower than the assumed 30%.

 

https://blog.revolutionanalytics.com/2013/01/chances-of-making-an-nfl-field-goal.html

1 minute ago, uticaclub said:

We beat the Jets & Titans because they missed FGs. It's just coming full circle in the second half of the season, now it's our time to lose games because of missed FGs.

No one says we cant learn from the experience of our opponents.

[billsfan714]

I was watching the Steelers game and Tomlin going for 4th and 1 from his own 34 with the lead and thinking no way McD makes that call.   Steelers convert, go on a 8 min drive, eating the clock    Ballsy.

[DaBillsFanSince1973]

14 minutes ago, dorquemada said:

 

I'm  not sure there's a bigger question than 4th and 4 for the game.

 

Additionally, whatever the Bills' analytics team is working on, they're not doing a very good job.  Unless, of course, Daboll's super cute and clever play calling is a result of analytics.  For example, when facing the #30 run defense, make sure to pass 75% of the time, they'll never know what hit them!

it's really why I'm done with Daboll.

 

done.

[Miyagi-Do Karate]

26 minutes ago, IgotBILLStopay said:

(click on the decision tree picture below to view a bigger version of the image)

 

McD admitted that they talked about whether to go for it on 4th and 4 or attempt the 53 yarder. Obviously they decided on the latter. I am revisiting that decision in this post since it appears very strange from a basic analytics perspective. That decision should be made based on 6 probability assumptions. 

 

1. The probability that Hauschka makes the 53 yarder (he had missed his previous 3 from 50+ yards and missed a 34 yarder earlier in the game - but holds the record for consecutive makes of 50+ yards)

2. The probability that the Bills convert a 4th and 4

3. The probability that the Bills score a TD in 15 seconds with a TO after converting the 4th down

4. The probability that the Hauschka makes a shorter field goal after the conversion on 4th and 4 

5. The probability that Bills win in OT

6. The Browns with a TimeOut and 22 seconds can get in position for a FG and make it

 

Here are my estimates:

 

1. 30% 2. 46% (https://blog.minitab.com/blog/the-statistics-game/calculating-the-probability-of-converting-on-4th-down)  3. 30% 4. 70% 5. 50% 6. 10%

 

 

That puts the win probability at 13.5 % for the 53 yard try  vs. 25.3% for going for it). The probability of a successful 53 yarder has to be close to 60% for the Bills to be comfortable going for it even keeping other probabilities same (not a given - if one has a 60% confidence in a Hauschka 53 yarder - the probability of a make from a shorter distance will be much higher than the assumed 70% - that will still suggest going for it on 4th down ). Listen - I continue to believe we lucked out in having a great coach who is able to prepare the players well and motivate them to play for each other. That being said, the decision to kick a 53 yarder appears quite short-sighted.

 

Dont the Bills have a basic analytics guy to advise them?

this is great stuff! 

[row_33]

it was 3rd and 4 with just under a minute on the clock and a time out in hand

 

and they wasted 20 or so seconds and stupidly threw long

 

this can't go on, hopefully lessons were learned for everyone involved

 

[dorquemada]

4 minutes ago, DaBillsFanSince1973 said:

it's really why I'm done with Daboll.

 

done.

 

His play calling manages to be both ineffective at moving the ball, situationally terrible (can't figure out how to get a first to burn clock to save his life), but also a terrible, boring grind to watch.   I don't even know how he does it.  A 12 year old mashing buttons in Madden would call a better game

[GG]

16 minutes ago, Buffalo Barbarian said:

 

they were 0% on 4th down conversions earlier in the game. What do the analytics say about that ??

 

 

 Even more of a reason to go for it because the odds are lower for being 0-3 on 4th down tries.

[IgotBILLStopay]

15 minutes ago, Buffalo Barbarian said:

 

they were 0% on 4th down conversions earlier in the game. What do the analytics say about that ??

 

 

If you dont have any faith in converting a 4 yard play, why kick a field goal and try to win with that offense in OT?

The lower your faith in converting a 4th down - the lower will be your probability of winning in OT even if a FG is successful, no?

[Da webster guy]

This is a good breakdown.   I don't see our HC or OC as particularly cerebral guys, probably just go for it on gut feeling.    HausMoney has won several games for us with long bombs, got us in the playoffs by stealing two games with fantastic, clutch kicks.   

 

Facing 4th and 4, I'd have trotted him out too, simply based on his career with us, not his recent struggles.  It's not a 60 yarder we're asking him to hit, its 20 yards longer than an extra point.

 

Another interesting analytic would be average rushing yards gained on 3rd down vs a 5 man front in late game situations where it's assumed you'll be passing.  Andy Reid does this all the time and it works like a charm.   Obvious passing down, less than a minute to go, one (or zero) timeouts.  Spread the field, keep your running back in as a blocker on the play before (we'd prob use Gore in this scenario, they'd think he was there to protect) and then on 3rd and 4 you run a draw play up the middle.   

 

It may not get you into the end zone, but their pass rushers are convinced its a pass and already upfield on the outside and you only have 3 defenders in the entire middle of the field.   Ten yard blast, first down on the 20, run up and spike the ball with 20 seconds left.  Two more chances for to try and score TD but your odds of hitting a FG just went WAY up and you get more chances with a new set of downs.

 

Teams freak out about clock and abandon the run when sometimes it's the perfect call.  Half the passes you complete the guy ends up getting tackled in bounds anyway.  Wish we had some more innovation and experience with game management on our staff.

 

 

[TheFunPolice]

so many coaches can design great systems, coach up their players, call a great game, then totally fall apart when it comes to situational awareness.

 

That's the biggest advantage NE has had for years now. They almost never lose, and when they do it's either to ex Ne assistants (TEN and DET last year) or other long-tenured coaches who tend to be somewhat better than average at this stuff.

 

Andy Reid is the most famous example of this, literally pissing away 1-2 wins a year with terrible clock and situation management. McDermott has been ok, but that end of game sequence was horrible. 

[BananaB]

I understand trying to kick the FG, what I don’t understand is that when we got to FG range there seemed to be no more urgency to go for the win..... Good chance we are gonna get a better FG opportunity if we a atleast tried to play for the TD. It was gutless in my opinion. 

Edited November 11, 2019 by BananaB

[MJS]

The team had already failed on fourth down earlier. You have to take the pulse of your own team, not just follow aggregated probability data. We have an offense that is ranked poorly. It's much different for a good or even average offense.

[vincec]

34 minutes ago, Mark80 said:

30% chance they score a td with 15 seconds left seems way high to me.

I was thinking the same thing. The Bills aren't the Saints. Even 20%, as subsequently suggested, seems very optimistic given what had happened up to that point in the game. With 15 seconds you basically get 2 plays before you have to kick the FG. If I had to guess, I'd say the chances of Buffalo scoring a TD from the 36 yard line in two plays with the Browns guarding the end zone are less than 5%.

Edited November 11, 2019 by vincec

[IgotBILLStopay]

1 minute ago, MJS said:

The team had already failed on fourth down earlier. You have to take the pulse of your own team, not just follow aggregated probability data. We have an offense that is ranked poorly. It's much different for a good or even average offense.

Ok let us concede this was McD's mindset. Let us say Hauschka makes the FG. How was McD planning to win in OT with that mindset about  the offensive pulse. Was he hoping for pick 6 or a turnover and have the defense win it for us?

[dorquemada]

2 minutes ago, MJS said:

The team had already failed on fourth down earlier. You have to take the pulse of your own team, not just follow aggregated probability data. We have an offense that is ranked poorly. It's much different for a good or even average offense.

 

They just stacked bad decisions.  Rather than a quick out to Beasley, or even one over the middle for a first, throw an ultra low probability pass to Brown, then have to kick from the full distance

[row_33]

Just now, IgotBILLStopay said:

Ok let us concede this was McD's mindset. Let us say Hauschka makes the FG. How was McD planning to win in OT with that mindset about  the offensive pulse. Was he hoping for pick 6 or a turnover and have the defense win it for us?

 

needs to go to a remedial class of How to Close an NFL Win Like  a Boss

 

that was pitiful, a learning situation hopefully

 

[Hapless Bills Fan]

41 minutes ago, IgotBILLStopay said:

(click on the decision tree picture below to view a bigger version of the image)

 

McD admitted that they talked about whether to go for it on 4th and 4 or attempt the 53 yarder. Obviously they decided on the latter. I am revisiting that decision in this post since it appears very strange from a basic analytics perspective. That decision should be made based on 6 probability assumptions. 

 

1. The probability that Hauschka makes the 53 yarder (he had missed his previous 3 from 50+ yards and missed a 34 yarder earlier in the game - but holds the record for consecutive makes of 50+ yards)

2. The probability that the Bills convert a 4th and 4

3. The probability that the Bills score a TD in 15 seconds with a TO after converting the 4th down

4. The probability that the Hauschka makes a shorter field goal after the conversion on 4th and 4 

5. The probability that Bills win in OT

6. The Browns with a TimeOut and 22 seconds can get in position for a FG and make it

 

Here are my estimates:

 

1. 30% 2. 46% (https://blog.minitab.com/blog/the-statistics-game/calculating-the-probability-of-converting-on-4th-down)  3. 30% 4. 70% 5. 50% 6. 10%

 

 

That puts the win probability at 13.5 % for the 53 yard try  vs. 25.3% for going for it). The probability of a successful 53 yarder has to be close to 60% for the Bills to be comfortable going for it even keeping other probabilities same (not a given - if one has a 60% confidence in a Hauschka 53 yarder - the probability of a make from a shorter distance will be much higher than the assumed 70% - that will still suggest going for it on 4th down ). Listen - I continue to believe we lucked out in having a great coach who is able to prepare the players well and motivate them to play for each other. That being said, the decision to kick a 53 yarder appears quite short-sighted.

 

Dont the Bills have a basic analytics guy to advise them?

 

 

I think the overall structure and conclusion is sound, but I'm unclear on where you got 70% fail probability for the 53 yards fieldgoal try?

Career, Haushka is 62.8% from 50+ yards.  But, he is 0 for 3 so far this season and IIRC he had a couple misses from >50 yds at the end of last season (?).  So I'm thinking that fail probability is probably higher than 70% yesterday.

 

It's telling that earlier in the game, when we had plenty of time to score more points, the Bills took 2 TOD rather than attempt a 53- and a 56-yd fieldgoal.

To me that says they know 53 yds is wishful thinking for Haushka right now.  And he'd also missed from 34 yards earlier that day!

 

[SectionC3]

29 minutes ago, H2o said:

We have a winner.  Also, FWIW, analytics doesn’t account for conditions on the date and time of the kick.  The Bills should have known Hauschka’s max based on pregame testing.  I can tell you for the Redskins it was 49 toward the tunnel end based on the fact that he kissed the crossbar short from 50.  How it was 9 feet further this week on grass and on what looked to be a somewhat slick field with no wind advantage is beyond me. 
 

 Morals of the story: Hauschka is nearing the end.  The play call on third and four was stupid.  And we deserved to lose for botching the sequence at the end.  

1 minute ago, Hapless Bills Fan said:

 

 

I think the overall structure and conclusion is sound, but I'm unclear on where you got 70% fail probability for the 53 yards fieldgoal try?

Career, Haushka is 62.8% from 50+ yards.  But, he is 0 for 3 so far this season and IIRC he had a couple misses from >50 yds at the end of last season (?).  So I'm thinking that fail probability is probably higher than 70% yesterday.

 

It's telling that earlier in the game, when we had plenty of time to score more points, the Bills took 2 TOD rather than attempt a 53- and a 56-yd fieldgoal.

To me that says they know 53 yds is wishful thinking for Haushka right now.  And he'd also missed from 34 yards earlier that day!

 

 

They do.  Watch the warmups about half an hour before the game.  All of this stuff is measured.  

Edited November 11, 2019 by SectionC3

[DasNootz]

Add in the probability of a game ending in a tie after OT and the numbers probably become slightly closer.  My gut is that the “right” call was to go for it based on Hauschka’s performance in preseason and this season.  

[JESSEFEFFER]

Both kicks looked like mishits to me compared to the Xpts he made going the same direction.  Sort of like my tee shots, it's tough to know anything else about the attempt when the ball is struck poorly.

[Numark3]

Even your made up statistics results in only 11% swing.  Which is pretty darn close.

[MJS]

13 minutes ago, dorquemada said:

 

They just stacked bad decisions.  Rather than a quick out to Beasley, or even one over the middle for a first, throw an ultra low probability pass to Brown, then have to kick from the full distance

And that's the biggest problem, I feel like. Why have we been passing it deep on 3rd and short? On some of those they didn't even have short crossing routes as an option.

[IgotBILLStopay]

21 minutes ago, vincec said:

I was thinking the same thing. The Bills aren't the Saints. What is this based on?

 

Successful conversrion of 4th and 4, we are at the 30 yard line with a 1st and 10. The average NFL chance of getting another first down after a 1st and 10 is 66% (read there - I am sure google can provide those). Let us start with 55% for us to move into the red zone (I am ignoring that we can win by scoring a TD).

 

The chance of a TD with a first and 10 from the redzone is 55% plus. https://www.footballoutsiders.com/stat-analysis/2013/closer-look-touchdowns-red-zone

 

55% times 55% gives you 30 %. But as I said in an earlier post - if you believe the chance of a successful conversion are only 20%, the win probability only drops to 22.08%  and is still significantly higher than win probability with kicking a 53 yarder. And if your faith in offense is lower than the above suggested numbers, that will come back to bite you even in the 53 yard attempt part of the decision tree.

13 minutes ago, Hapless Bills Fan said:

 

 

I think the overall structure and conclusion is sound, but I'm unclear on where you got 70% fail probability for the 53 yards fieldgoal try?

Career, Haushka is 62.8% from 50+ yards.  But, he is 0 for 3 so far this season and IIRC he had a couple misses from >50 yds at the end of last season (?).  So I'm thinking that fail probability is probably higher than 70% yesterday.

 

It's telling that earlier in the game, when we had plenty of time to score more points, the Bills took 2 TOD rather than attempt a 53- and a 56-yd fieldgoal.

To me that says they know 53 yds is wishful thinking for Haushka right now.  And he'd also missed from 34 yards earlier that day!

 

Thanks for pointing that out. I was giving McD the benefit of doubt there by assuming a 70% fail probability. If the fail probability is greater (and objectively it is), that only makes the decision to kick appear even weirder. Assuming a 90% fail probability makes the win probability just 4.5 percent for the kicking decision.

Edited November 11, 2019 by IgotBILLStopay

[skibum]

It was the right call to make, and not even in the top ten worst play calls of the game. Hauschka is capable of making a 52 yarder, just not that one apparently. 

[Hapless Bills Fan]

6 minutes ago, IgotBILLStopay said:

Thanks for pointing that out. I was giving McD the benefit of doubt there by assuming a 70% fail probability. If the fail probability is greater (and objectively it is), that only makes the decision to kick appear even weirder. Assuming a 90% fail probability makes the win probability just 4.5 percent for the kicking decision.

 

That's about where I got to.  And yet, McDermott has full confidence and faith in Hauschka. 

[CaptnCoke11]

39 minutes ago, billsfan714 said:

I was watching the Steelers game and Tomlin going for 4th and 1 from his own 34 with the lead and thinking no way McD makes that call.   Steelers convert, go on a 8 min drive, eating the clock    Ballsy.

McDermott would be 1-8 with that Steelers team.  That’s the difference 

[2003Contenders]

I get the sense that McD and Daboll are NOT properly communicating on 3rd down plays when there is a potential to go for it on 4th (in the event that the 3rd down conversion is not met).

 

Maybe if Daboll knows that McD is willing to give him the green light to go for it on 4th down if we don't convert 3rd and 4, then maybe the play call is different. That is, maybe running it on 3rd and 4 at least picks up a couple of yards, even if the runner doesn't make it to the 1st down marker. Suddenly, 4th and 1 or 2 looks a lot better than 4th and 4, right?

[IgotBILLStopay]

9 minutes ago, Crayola64 said:

Even your made up statistics results in only 11% swing.  Which is pretty darn close.

1. I was trying to stack the deck in favor of the kicking decision - so the assumed probabilities are conservative. As pointed out by HaplessBillsFan earlier, the fail percentage on the 53 yarder is likely  greater than 70%. That will make the difference even larger.

2. Dont sneer at 11% my friend - coaches will die for an ex-ante one-play 11% increase in win probability.

6 minutes ago, skibum said:

It was the right call to make, and not even in the top ten worst play calls of the game. Hauschka is capable of making a 52 yarder, just not that one apparently. 

Capable? Yes. Probable? No.