Aaron Schatz (Football Outsiders) on the Bills schedule

[BillsVet]

Clearly Dave you are a hater.  

 

Why did you have to go rain on this parade? 

 

 

[SlimShady'sSpaceForce]

On 10/22/2019 at 4:57 PM, Bmbchiefs66 said:

Dvoa is very simple.to under stand

 

DVOA proves what exactly? 

 

Is it indicative of actually getting more wins? 

[JESSEFEFFER]

Some of the analysis is lazy, imo.  These analytics guys are starting to irritate me which is ironic because I am often accused of being too analytical. Of the four Titan missed kicks,  2 were from 50 and 53  (less than 50/50 to make both outdoors off a grass field) and 1 was blocked.  The Bills had the ball to close the game on a 50 yard drive with the Titans needing the ball back to have any chance to tie.  That is not directly tied to a loss.  Missing a kick at the end of a game to win is a direct tie.  The Jets missed kicks are no more flukey than 3 of the Bills' 4 turnovers in that game.  The Bills way outperformed the Jets and the Titans, by big margins except on the scoreboard.

[Hapless Bills Fan]

On 10/22/2019 at 4:15 PM, Kelly the Dog said:

One thing you cannot accuse Dave of is being a homer or leaning toward something because it is warm and fuzzy and says good things about the Bills. You should know that.

 

I'm not "accusing" Dave of anything.  I believe his point was he likes DVOA, and to inform me that DVOA is saying good stuff about Bills D so perhaps I should reconsider my reservations.

 

I don't roll that way, though.  If I have reservations about an "advanced stat" or whatever these things are called, I will have them whether they tell me the Bills QB is #9 or 10 overall at the end of the season (PFF, 2015) or whether they tell me he sucks.   Same with DVOA, reservations don't disappear because they said Bills D was #2 last year.

 

Like total QBR, I think DVOA is trying to boil too many things into one statistic, and usually when that's done, if one is allowed to look under the hood one sees that subjectivity and an equation with degrees of freedom unsupported by the statistical significance of the data it's using have gotten involved.

 

As a guy I once worked with quipped, if you use enough parameters, you can eventually make a curve fit the trace of an elephant, including the trunk and the tail.  Doesn't mean you can necessarily tell the difference between an elephant and a rhinoceros using your model.

 

18 hours ago, dave mcbride said:

For those who don't like Schatz, you'll love this quote! 

 

http://www.footballoutsiders.com/dvoa-ratings/2019/week-7-dvoa-ratings

Combine all three phases of the game, and the Bills hold steady at No. 21 in the DVOA ratings. That's not where we normally find a 5-1 team. How bad is it? The Bills are officially the worst 5-1 team in DVOA history.

 

If the territory isn't being well described by the map....maybe the map has a problem.  Or maybe they're right and the Bills will fall apart.

 

I'm hoping for the first, but time will tell.

Just off the top, here are two of my reservations about Schatz current analysis and prognostications?

1) From FO own explanation of DVOA/DYAR in the "Issues" section: DVOA is limited by what’s included in the official NFL play-by-play or tracked by the Football Outsiders game charting project. Because we need to have the entire play-by-play of a season in order to compute DVOA and DYAR.

FO admits they look for sufficient statistical significance by using all the plays made during a season.  But they are now using it partway through a season to make prognostications.  OK - are they sure that same model that seems predictive with all the data of a season, is still statistically significant and valid 1/3 of the way through?  Are they filling in by rolling in data from the previous season (this is, as I understand it, what 538 does)?  Since they don't tell anyone exactly what they're doing or what their model is, we really don't know.

2) He says "We then expand upon that basic idea with a more complicated system of “success points,” improved over the past few years with a lot of mathematics and a bit of trial and error."  

Like the rattle of a rattlesnake warns there's a venomous snake in the vicinity, the bolded are "buzz words" that warn the scientist that an empirical model is in use.  What that means is they're messing with the weighting and perhaps sticking in extra variables in ways that aren't directly linked to actual data.  One problem with such models is the aforementioned one of statistical significance vs DOF.  Another other well-known problem is that of multiple minima.  Maybe they've sorted all that, or maybe they haven't - like draft board pundits, they don't suffer much consequence if their data doesn't shake out, and in any event, they aren't inviting the world in to rummage about in their method and see what they do.

 

OK, I think I'm done now.

 

On 10/22/2019 at 4:33 PM, dave mcbride said:

With "maybe," I thought you were implicitly questioning my claim that he wasn't saying they were bad or mediocre.  He clearly likes the Bills defense (just as his system hated it under Rex). I wasn't commenting on your skepticism about DVOA.

 

Thanks for the explanation, understand now.  And I've found a point of agreement with FO system since I hated our D under Rex. ?

 

What the eyeball test and my preferred stats tell me is that our offensive output will probably not sustain winning if we don't improve it.  Allen's ratio of TD to turnovers is too high (even if one includes rushing TDs).  In 2 most recent games, he has tamed that down, but at the cost of some passing productivity which is now marginal.  He's a young QB and seems to be continuing to develop, so we just have to see where he goes.   If he keeps the TO's tamed but completes even 2-4 extra passes per game, we should be OK.  I think the kid has a level head and a work ethic to go with his physical skills, so I'm officially Hopeful, where Allen is concerned.  But I've been Hopeful before on a Bills QB, and been wrong.  ?‍♀️

 

What the FO model almost certainly does is say "as the teams are now so shall they be".   That's an assumption that doesn't account for injuries (which can be assumed to be random) or in-season changes in a player's productivity (good or bad).

 

[dollars 2 donuts]

On 10/22/2019 at 4:12 PM, Royale with Cheese said:

 

Well yeah...I was just pointing this out because when we were faced against a top offense, we were very good.

So we aren't just showing dominance against bad offenses...

 

 

Damn straight, Royale.

[dollars 2 donuts]

On 10/23/2019 at 8:48 AM, eball said:

 

Yep, and that's why I'm not giving up hope to win the division.  Wouldn't it be incredible if that W16 game @ Foxboro actually meant winning or losing the East?

 

eball, I mean seriously, what if ("if" everybody) both teams are 13-1 when the Bills go there?  There will not be any "divisional tomorrow", meaning the winner of that game, regardless of how they win it, just needs to beat their final opponent to lock up home field advantage all the way through playoffs.

 

I will take ALL of the wins we get.  I don't need moral losses or comeback to Earth games.

 

Additionally, and staying with this point, next year's team could technically be better than this year's, but might only get to 12-4.

 

...strike while the iron is hot.

Edited October 24, 2019 by dollars 2 donuts

[dave mcbride]

2 hours ago, Hapless Bills Fan said:

 

I'm not "accusing" Dave of anything.  I believe his point was he likes DVOA, and to inform me that DVOA is saying good stuff about Bills D so perhaps I should reconsider my reservations.

 

I don't roll that way, though.  If I have reservations about an "advanced stat" or whatever these things are called, I will have them whether they tell me the Bills QB is #9 or 10 overall at the end of the season (PFF, 2015) or whether they tell me he sucks.   Same with DVOA, reservations don't disappear because they said Bills D was #2 last year.

 

Like total QBR, I think DVOA is trying to boil too many things into one statistic, and usually when that's done, if one is allowed to look under the hood one sees that subjectivity and an equation with degrees of freedom unsupported by the statistical significance of the data it's using have gotten involved.

 

As a guy I once worked with quipped, if you use enough parameters, you can eventually make a curve fit the trace of an elephant, including the trunk and the tail.  Doesn't mean you can necessarily tell the difference between an elephant and a rhinoceros using your model.

 

 

If the territory isn't being well described by the map....maybe the map has a problem.  Or maybe they're right and the Bills will fall apart.

 

I'm hoping for the first, but time will tell.

Just off the top, here are two of my reservations about Schatz current analysis and prognostications?

1) From FO own explanation of DVOA/DYAR in the "Issues" section: DVOA is limited by what’s included in the official NFL play-by-play or tracked by the Football Outsiders game charting project. Because we need to have the entire play-by-play of a season in order to compute DVOA and DYAR.

FO admits they look for sufficient statistical significance by using all the plays made during a season.  But they are now using it partway through a season to make prognostications.  OK - are they sure that same model that seems predictive with all the data of a season, is still statistically significant and valid 1/3 of the way through?  Are they filling in by rolling in data from the previous season (this is, as I understand it, what 538 does)?  Since they don't tell anyone exactly what they're doing or what their model is, we really don't know.

2) He says "We then expand upon that basic idea with a more complicated system of “success points,” improved over the past few years with a lot of mathematics and a bit of trial and error."  

Like the rattle of a rattlesnake warns there's a venomous snake in the vicinity, the bolded are "buzz words" that warn the scientist that an empirical model is in use.  What that means is they're messing with the weighting and perhaps sticking in extra variables in ways that aren't directly linked to actual data.  One problem with such models is the aforementioned one of statistical significance vs DOF.  Another other well-known problem is that of multiple minima.  Maybe they've sorted all that, or maybe they haven't - like draft board pundits, they don't suffer much consequence if their data doesn't shake out, and in any event, they aren't inviting the world in to rummage about in their method and see what they do.

 

OK, I think I'm done now.

 

 

Thanks for the explanation, understand now.  And I've found a point of agreement with FO system since I hated our D under Rex. ?

 

What the eyeball test and my preferred stats tell me is that our offensive output will probably not sustain winning if we don't improve it.  Allen's ratio of TD to turnovers is too high (even if one includes rushing TDs).  In 2 most recent games, he has tamed that down, but at the cost of some passing productivity which is now marginal.  He's a young QB and seems to be continuing to develop, so we just have to see where he goes.   If he keeps the TO's tamed but completes even 2-4 extra passes per game, we should be OK.  I think the kid has a level head and a work ethic to go with his physical skills, so I'm officially Hopeful, where Allen is concerned.  But I've been Hopeful before on a Bills QB, and been wrong.  ?‍♀️

 

What the FO model almost certainly does is say "as the teams are now so shall they be".   That's an assumption that doesn't account for injuries (which can be assumed to be random) or in-season changes in a player's productivity (good or bad).

 

Good post. What I think that the system really doesn't really account for is coaching acumen in close games (and there are always a lot of close games -- just ask Belichick). A really good team can dominate most games and still go 8-8. I don't think that who wins close games in the aggregate comes down to a series of coin flips (in the probabilistic sense). For instance, the 2017 Bills were a bad team talent wise and still went 9-7 because they were situationally good in a lot of close games while getting blown in games in which they were overmatched. I even think that their record last year was a product of coaching because in reality they could have easily been a 3-13 team given the abysmal talent on offense. In contrast, Dan Quinn (for instance) has overseen some excellent teams in terms of talent and never gotten beyond 11-5 (plus he lost to an overmatched Bills team in 2017 by blowing a playcall near the goal line late -- the Bills only had 10 men on the field and he decided to pass it on a fairly easy run play). Atlanta has lost a bunch of close games in the past couple of seasons, and I think any reasonable observer would say that they probably have had more talent than the Bills. That has to be on coaching. Indeed, losing to NE in a game they had in the bag in the SB says it all. Passing from the 20 yard line instead of running and settling for a FG that would have won the game (and with a kicker who was perfect up to that point) was a TERRIBLE decision.  

 

Another example: SD handing it off last week to Melvin Gordon on the final play after he fumbled and recovered it the previous play (nullifying a TD). He then fumbles it AGAIN and possibly loses it (TN got the call, but it was murky), but even if he hadn't fumbled it, he hadn't gotten in, and since they were out of timeouts,the game would have ended despite there being about 20-25 seconds on the clock when the play started. You have got to try a couple of passes there and only run for it when that fails. It'll give you three shots instead of one, which increases your odds of actually punching it in. And moreover, handing it off to a guy who was shaky on the previous play probably wasn't wise. SD was the better team in that game, and they lost. 

 

With regard to the Bills this season, I do think he's sort of right about the flukiness of the kicking, but then again the Jets' coaching staff made an organizational decision to let a solid kicker go before the season began and settle for a bad one. The TN game was genuinely fluky, though, because Santos had been a decent kicker. 

Edited October 24, 2019 by dave mcbride

[RoyBatty is alive]

 top offense as in the Dolphins, LOL

[Hapless Bills Fan]

36 minutes ago, dave mcbride said:

Good post. What I think that the system really doesn't really account for is coaching acumen in close games (and there are always a lot of close games -- just ask Belichick). A really good team can dominate most games and still go 8-8. I don't think that who wins close games in the aggregate comes down to a series of coin flips (in the probabilistic sense). For instance, the 2017 Bills were a bad team talent wise and still went 9-7 because they were situationally good in a lot of close games while getting blown in games in which they were overmatched. I even think that their record last year was a product of coaching because in reality they could have easily been a 3-13 team given the abysmal talent on offense. In contrast, Dan Quinn (for instance) has overseen some excellent teams in terms of talent and never gotten beyond 11-5 (plus he lost to an overmatched Bills team in 2017 by blowing a playcall near the goal line late -- the Bills only had 10 men on the field and he decided to pass it on a fairly easy run play). Atlanta has lost a bunch of close games in the past couple of seasons, and I think any reasonable observer would say that they probably have had more talent than the Bills. That has to be on coaching. Indeed, losing to NE in a game they had in the bag in the SB says it all. Passing from the 20 yard line instead of running and settling for a FG that would have won the game (and with a kicker who was perfect up to that point) was a TERRIBLE decision.  

 

Another example: SD handing it off last week to Melvin Gordon on the final play after he fumbled and recovered it the previous play (nullifying a TD). He then fumbles it AGAIN and possibly loses it (TN got the call, but it was murky), but even if he hadn't fumbled it, he hadn't gotten in, and since they were out of timeouts,the game would have ended despite there being about 20-25 seconds on the clock when the play started. You have got to try a couple of passes there and only run for it when that fails. It'll give you three shots instead of one, which increases your odds of actually punching it in. And moreover, handing it off to a guy who was shaky on the previous play probably wasn't wise. SD was the better team in that game, and they lost. 

 

With regard to the Bills this season, I do think he's sort of right about the flukiness of the kicking, but then again the Jets' coaching staff made an organizational decision to let a solid kicker go before the season began and settle for a bad one. The TN game was genuinely fluky, though, because Santos had been a decent kicker. 

 

Haha we're going to have a mutual admiration society going down here - likewise, good post.

 

You're right about "a lot of close games", and about coaching making a difference in them.   Another coach who seems to make poor situational decisions when it counts most would be Andy Reid.

 

Another factor DVOA mid-season clearly can't account for: changes in personnel, either due to injury or due to a player just stepping up and improving as the season goes on or, a player falling off as the season goes on. 

 

To twist the old phrase, it's about both the X's and O's AND the Jimmy's and Joe's.

 

The TN game field goal was not that fluky actually.  4 FG: 1 blocked.  That happens, not his fault.  Two, though, were from beyond 50 yds and excluding this year, Santos was 7 of 14 from 50 yards or more.  That's just his limit, he's a 50% at that distance.  The short one was a fluke for a previously good kicker.  So 1:4 fluke maybe 2:4 if you count the block as a fluke.

 

 

[buffswagger018]

3 hours ago, Hapless Bills Fan said:

 

Haha we're going to have a mutual admiration society going down here - likewise, good post.

 

You're right about "a lot of close games", and about coaching making a difference in them.   Another coach who seems to make poor situational decisions when it counts most would be Andy Reid.

 

Another factor DVOA mid-season clearly can't account for: changes in personnel, either due to injury or due to a player just stepping up and improving as the season goes on or, a player falling off as the season goes on. 

 

To twist the old phrase, it's about both the X's and O's AND the Jimmy's and Joe's.

 

The TN game field goal was not that fluky actually.  4 FG: 1 blocked.  That happens, not his fault.  Two, though, were from beyond 50 yds and excluding this year, Santos was 7 of 14 from 50 yards or more.  That's just his limit, he's a 50% at that distance.  The short one was a fluke for a previously good kicker.  So 1:4 fluke maybe 2:4 if you count the block as a fluke.

 

 

You could be right....*shoulder shrug.... We shall see

[billsredneck1]

On 10/22/2019 at 4:15 PM, Mrbojanglezs said:

Good. I like easy schedules. Why would we want a hard schedule? 

i don't buy that crap this early. there's always at least a couple of teams that get on a roll out of nowhere.

[dave mcbride]

@Hapless Bills Fan - circling back on this. 

 

https://www.footballoutsiders.com/stats/teamdef/2019

 

through the miami game, FO has the bills’ D ranked 12th, dropped from 3rd the previous week and despite still being third in both points and yards allowed. Given this Eagles game, it looks like their assessment of who they really are on D is more or less correct. 

[Hapless Bills Fan]

Just now, dave mcbride said:

@Hapless Bills Fan - circling back on this. 

 

https://www.footballoutsiders.com/stats/teamdef/2019

 

through the miami game, FO has the bills’ D ranked 12th, dropped from 3rd the previous week and despite still being third in both points and yards allowed. Given this Eagles game, it looks like their assessment of who they really are on D is more or less correct. 

 

Time will tell, but when you get your butt whipped on D like we did today, there’s really nothing to say in defense of our defense