2023 Defensive Save % Rankings (Like Baseball)

[Mikie2times]

What motivated this was the desire to quantify the late game collapses. We all talk about how we struggle to close out games but then I wonder what is normal? What is struggling? I hate that we don’t have context.

 

Even with advanced analytics very little is available. So while not perfect, this is my first swing at such a stat.  

 

I looked at this the same way baseball looks at closing out games. I did not care about the outcome, only if a lead was lost and when it was lost. This was the criteria I used (understandably imperfect).

 

1.  Did the qualifying team have a one possession lead (8 points or less) in the game with 5 minutes or less or in OT => IF YES go to criteria 2

2. Did the opponent have a chance to possess the ball in the final 5 minutes => IF YES go to criteria 3

3. Did the opponent have at least 30 seconds down by 3 points or less or 1 minute down by a max of 8 points when they possessed the ball? If YES the situation qualifies for a SAVE

 

You can’t get a Save for protecting or blowing a tie. It's just like baseball, you can only have a qualifying save opportunity by protecting a lead that meets this criteria. 

 

You can also qualify for multiple saves in one game (this was rare, but Buffalo has done this a few times). As quick example of one, the Eagles had a qualifying save opportunity on our final drive of regulation. They had the lead, it met the criteria outlined. They blew the save when Gabe Davis scored his TD with 1:52 left. Since that score then gave Buffalo a lead, the Bills proceeded to blow a save when the Eagles tied the game at the end of regulation. The other team does not have to take the lead, they can tie game to blow a save, you just have to lose the lead. The Bills then blew another Save in OT as they again had the lead only to allow a game winning score.  This one game resulted in 3 Blown saves. 2 for Buffalo and 1 for the Eagles. While more than 1 blows save per game is rare, it can happen. It's the NFL.

 

Edited* I was asked to include seasons prior to 2020, so now this includes those seasons, editing to reflect that  

 

The Bills results are as follows:

2017: 2 of 3 66% Save Rate

2018: 1 of 4  25% Save Rate

2019: 4 of 7 57% Save Rates

2020: 2 for 4 50% Save Rate

2021: 0 for 3 0% Save Rate

2022: 2 for 4 50% Save Rate

2023: 3 for 8 37.5% Save Rate

Total: 14 for 33 42.4% Save Rate

 

So if you're saying, hey, does that mean if we have an 8 point lead or less under 5 minutes and our opponent touches the ball we will preserve that lead 42.4% of the time? Yes. That's what it means.

 

Below is all of 2023 by team, please jeep in mind the 2023 summary is more for the aggregate to compare against us. The individual team data is just what comes with it. 

 

Removing Buffalo, the NFL average is 51.7% on 141 save chances in 2023. Which seems low, but again, you can have two blown saves in a game. You can't have two successful saves. 

 

This is stack ranked best to worst. Some interesting teams in here but more data is required.

 

Click on the image to enlarge it. 

 

 

Edited December 28, 2023 by Mikie2times

[ClemsonBills]

Yikes. 

[Billy Claude]

I appreciate the effort but the problem with stats like this is that there is just not enough data points in a season to make it statistically meaningful.

[Mikie2times]

1 minute ago, Billy Claude said:

I appreciate the effort but the problem with stats like this is that there is just not enough data points in a season to make it statistically meaningful.

While I don't disagree, we could say the same thing about wins and losses

[Billy Claude]

3 minutes ago, Mikie2times said:

While I don't disagree, we could say the same thing about wins and losses

The average save attempts per team looks to be about six.   If you compare it to won/loss records, that would be similar to randomly taking a six game stretch of each team's season and using it for the final standings.  The standings would definitely correlate with the overall season standings but there would also be a lot of differences.

[Mikie2times]

1 minute ago, Billy Claude said:

The average save attempts per team looks to be about six.   If you compare it to won/loss records, that would be similar to randomly taking a six game stretch of each team's season and using it for the final standings.  The standings would definitely correlate with the overall season standings but there would also be a lot of differences.

How does Buffalo perform? How does everybody else perform. I didn't care so much about how each individual team performed. I was looking for the aggregate.  Buffalo has 19 samples. Everybody else has 141 samples. It's likely pretty safe to say the NFL average will fall between 50-55%. As that sample is decent in size and that % is somewhat logical. As for Buffalo, it is what it is right now. It's like W/L. 19 isn't enough to say forever but it's more than a full year of games which we find relevant. 

[Doc Brown]

I appreciate the effort but way too small of a sample size.  Now if you really feel ambitious start from McDermott's first year and see how the defense has performed the last five and a half seasons to gain anything meaningful.

[Mikie2times]

Just now, Doc Brown said:

I appreciate the effort but way too small of a sample size.  Now if you really feel ambitious start from McDermott's first year and see how the defense has performed the last five and a half seasons to gain anything meaningful.

The sample size is 19 from 2000 on and 141 in 2023 so far. If you want me to add other years that's fine. I went with 2020 because most people make sure you don't include non prime Allen years for any Bills analysis. 

[WotAGuy]

I appreciate the effort this took and the motivation to share it with the board.  These types of posts add quality to the board content. 
 

While the approach has its limitations, it does provide the answer you were looking for - you can see the Bills are in the lower third at save percentage.
 

That seems consistent with what we have seen and felt this season. 

[Billy Claude]

5 minutes ago, Mikie2times said:

How does Buffalo perform? How does everybody else perform. I didn't care so much about how each individual team performed. I was looking for the aggregate.  Buffalo has 19 samples. Everybody else has 141 samples. It's likely pretty safe to say the NFL average will fall between 50-55%. As that sample is decent in size and that % is somewhat logical. As for Buffalo, it is what it is right now. It's like W/L. 19 isn't enough to say forever but it's more than a full year of games which we find relevant. 

 

I was just looking at your data for this year.  I agree that your 5 year data for the Bills supports that the Bills D has been bad at "save" situations.

[Doc Brown]

4 minutes ago, Mikie2times said:

The sample size is 19 from 2000 on and 141 in 2023 so far. If you want me to add other years that's fine. I went with 2020 because most people make sure you don't include non prime Allen years for any Bills analysis. 

You mean from 2020 on and I again appreciate the effort.  I would be interested from 2017 to 2019 to see if it's significantly different than the last three and a half years.  I remember the Bills being better early in McDermott's tenure but I could be wrong.

[Mikie2times]

7 minutes ago, Doc Brown said:

You mean from 2020 on and I again appreciate the effort.  I would be interested from 2017 to 2019 to see if it's significantly different than the last three and a half years.  I remember the Bills being better early in McDermott's tenure but I could be wrong.

That was a horrifying request. The mind really does forget some of those. 2019 was pretty good, but that was the Houston game.  Below average still. but the sub 40% was damning. Sort of interesting how this years team has put him in so many of these spots. The most of any year in his career and double most years. This includes the playoffs. 

 

 

[Doc Brown]

Looks about right.  About league average the first three years and then that pesky elite gunslinger came along.

[Mikie2times]

4 minutes ago, Doc Brown said:

Looks about right.  About league average the first three years and then that pesky elite gunslinger came along.

So what narrative do you buy into? 13 seconds actually did mean something? Just randomness and no worries? Something else?

[LABILLBACKER]

22 minutes ago, Mikie2times said:

So what narrative do you buy into? 13 seconds actually did mean something? Just randomness and no worries? Something else?

13 seconds was so embarrassing it should count as 2 lost attempts.

[Mikie2times]

2 minutes ago, LABILLBACKER said:

13 seconds was so embarrassing it should count as 2 lost attempts.

It was 2 blown saves. One on the Tyreek TD and one on the FG for OT.  KC also had 2 blown saves in that game. It was the only game I saw out of all of them that had 4 blown saves. 

[RyanC883]

This is great stuff!

 

Curious about number 3.  How was this metric chosen for qualifying a save?  Is this basically the same as saying that a “save” is preventing the other team from eliminating its deficit (of 8 or fewer points) in the qualifying minutes set forth in metric 1? 

[vincec]

48 minutes ago, Mikie2times said:

That was a horrifying request. The mind really does forget some of those. 2019 was pretty good, but that was the Houston game.  Below average still. but the sub 40% was damning. Sort of interesting how this years team has put him in so many of these spots. The most of any year in his career and double most years. This includes the playoffs. 

 

 

I don’t see how you can add up long stretches of seasons and get anything coherent. Probably 90% of Buffalo’s defensive personnel and the play caller are different from 2017 to 2023. So how can you add those numbers together to get a picture of anything meaningful?

[Mikie2times]

9 minutes ago, vincec said:

I don’t see how you can add up long stretches of seasons and get anything coherent. Probably 90% of Buffalo’s defensive personnel and the play caller are different from 2017 to 2023. So how can you add those numbers together to get a picture of anything meaningful?

My general thought is McD being the common denominator. Being a defensive based HC, I think it's fair to roll this to him. If not, probably fair to roll it to the people he had in charge and then roll it to him. I stated from the outset, no perfect way to do this, but if you want to know how many times we had a lead under 5 minutes and lost that lead, this would probably be the way to do it. I would think......

Edited December 28, 2023 by Mikie2times

[SinceThe70s]

2 hours ago, Mikie2times said:

What motivated this was the desire to quantify the late game collapses. We all talk about how we struggle to close out games but then I wonder what is normal? What is struggling? I hate that we don’t have context.

 

Even with advanced analytics very little is available. So while not perfect, this is my first swing at such a stat.  

 

I looked at this the same way baseball looks at closing out games. I did not care about the outcome, only if a lead was lost and when it was lost. This was the criteria I used (understandably imperfect).

 

1.  Did the qualifying team have a one possession lead (8 points or less) in the game with 5 minutes or less or in OT => IF YES go to criteria 2

2. Did the opponent have a chance to possess the ball in the final 5 minutes => IF YES go to criteria 3

3. Did the opponent have at least 30 seconds down by 3 points or less or 1 minute down by a max of 8 points when they possessed the ball? If YES the situation qualifies for a SAVE

 

You can’t get a Save for protecting or blowing a tie. It's just like baseball, you can only have a qualifying save opportunity by protecting a lead that meets this criteria. 

 

You can also qualify for multiple saves in one game (this was rare, but Buffalo has done this a few times). As quick example of one, the Eagles had a qualifying save opportunity on our final drive of regulation. They had the lead, it met the criteria outlined. They blew the save when Gabe Davis scored his TD with 1:52 left. Since that score then gave Buffalo a lead, the Bills proceeded to blow a save when the Eagles tied the game at the end of regulation. The other team does not have to take the lead, they can tie game to blow a save, you just have to lose the lead. The Bills then blew another Save in OT as they again had the lead only to allow a game winning score.  This one game resulted in 3 Blown saves. 2 for Buffalo and 1 for the Eagles. While more than 1 blows save per game is rare, it can happen. It's the NFL.

 

The Bills results are as follows:

2020: 2 for 4 50% Save Rate

2021: 0 for 3 0% Save Rate

2022: 2 for 4 50% Save Rate

2023: 3 for 8 37.5% Save Rate

Total: 7 for 19 36.84% Save Rate

 

So if you're saying, hey, does that mean if we have an 8 point lead or less under 5 minutes and our opponent touches the ball we will preserve that lead 36.84% of the time? Yes. That's what it means.

 

Below is all of 2023 by team

 

Removing Buffalo, the NFL average is 51.7%. Which seems low, but again, you can have two blown saves in a game. You can't have two successful saves. 

 

This is stack ranked best to worst. Some interesting teams in here. 

 

Click on the image to enlarge it. 

 

 

 

The odd thing about the chart to me is that the two teams that currently look (to my eyes) like the class of their respective conferences (49ers and Ravens) are at the bottom of the chart with a combined 0-6.

[Mikie2times]

4 minutes ago, SinceThe70s said:

 

The odd thing about the chart to me is that the two teams that currently look (to my eyes) like the class of their respective conferences (49ers and Ravens) are at the bottom of the chart with a combined 0-6.

For sure. Again, the aggregate was really the only reason I have the team totals. I knew I would run multiple seasons against Buffalo to compare against the aggregate. As for observations with some of the other teams.

 

-The 49ers just haven't been in close games. Only one trial.

 

-The Ravens are a team I would be interested in maybe seeing more. If you recall, they blew like a 20 point lead to us in 2021 I believe. They blew a Wild Card lead to the Bengals last year. I think they might have some chronic issues if we go further back. 

 

-Green Bay, Seattle, Titans, Eagles, Chargers I have some curiosity. It would be nice to know who is the low in this sample over a few years and if any teams show patterns. I would think the Chargers would. 

 

- Pittsburgh, Tomlin has always had this magical ability to get the most out of his guys. Is this it? What is the driver? Closers on defense? Does this stretch multiple seasons?

[julian]

I like the effort here, I approve.

[Mikie2times]

2 hours ago, RyanC883 said:

This is great stuff!

 

Curious about number 3.  How was this metric chosen for qualifying a save?  Is this basically the same as saying that a “save” is preventing the other team from eliminating its deficit (of 8 or fewer points) in the qualifying minutes set forth in metric 1? 

The 30 second vs 1 minute threshold was just put in place to say, to consider this a blown save the other team needs adequate time to actually threaten you.The scoring threshold was put in place to determine "runners in scoring position" concept. In baseball you only get credit for a save if the other team is actually a threat. So this had to be one score. The time thresholds are arbitrary. We could say instead of 5 minutes, make it 7 minutes. Instead of a minimum 30 seconds for a FG make it 15 seconds. I tried to be on the side of a true blown save. A clear situation where a stop was required and it wasn't made. 

 

About 5 instances occurred where a blown save did occur under the time threshold of 30 seconds for a FG and 1 minute for a TD (13 seconds among them). In these rare examples I counted the blown save. As it was even worse in my mind than having adequate time. I did keep that concept uniform. In the other examples in which the save wasn't blown but it was below the time threshold it just wasn't counted at all. Hope that makes sense. As I plowed into this I had to keep creating nuances to account for the various situations.