Many know i love data, but it has to be relevant and from the real world.

QB Rate was the first HUGELY flawed data point, then espn QBR came along which was 70% better, which is fine.

As we see the spikes of oddity data happening here or there in a season, and fans just accepting shit data, and media accepting shit data and ignoring reality.

The bengals defense was a victim of this along with a victim of the bad players and key injuries, but the data is also flawed, right?

we gave up 58 points to the opposition while we were on offense. thats an outlier by a mile, the average is in the 15 pt range, so almost quadruple.

So im not here to say "see we were only the 25th worst defense" i dont care about that, i want to go far deeper than that.

Being in Denver the Bo Nox hype is off the rails, and the broncos have possibly the best defense i have seen since the seahawks SB runs?

bo nix has the easiest QB job in the nfl. and it made me start thinking even more about defensive ratings.

here are a few points i think are important in gauging a defense, its shortcomings, etc.

1. Obviously leaving out points given up when they werent on the field, offense and special teams, why isnt this already done?

2. factor what led to points being scored, a turnover, a flubbed punt, a turnover on downs that gives the ball to the other team in scoring position should be weighted. Like a pick thrown at your own 20 yardline, defense gives up 2 yards and a FG on the drive after, thats FLAWLESS defense, and their rating just dropped.

3. Weighting situational football, 3rd down rate giving up 4/4 on 3rd and 1 to tush pushes, vs giving up 4/4 on 3rd and 11 to blown plays or coverage show up the same on the data, this is foolish, is this data "vs probability"? like they give up 70% success of third downs that carry a 30% success rate in the league, they suck, vs "they give u p a 10% success rate on 3rd downs that carry a 80% success rate" they are great...

4. Scoring has to be king in any rating with a defense, which is why #1 is so important, like the QBR not weighing a 40 yard air td pass to tee the same as an andy dalton special dropping the ball into hawkins hands in the backfield he runs for 23 yards TD... some thing for defenses..

5. Lastly, field position over average, a good data point showing hey, this defense gives this offense a way better than average position on the field, showing the bend dont break repercussions on your offense, etc.

I think having some real data would change how some defenses are viewed, and players like Bo nix who are handed turnovers, great field position, and more opportunities than others do, and you do less with it..

this can go deep and insane levels which are unrealistic. the TE issues we had this season are a great example.

we dont have an issue covering TE's nearly as much as we had a pressure on the QB problem, we werent often getting diced up on tempo to TE's play after play.. we werent getting to the QB, they bought time, and the TE is both often covered by the weakest cover guy or in zone time lets you find a hole in zone, and WR are priority, LB safety etc often end up covering TE's.

and on QB's who can move in the pocket and buy time, those are the QB who cant play as much man coverage as backs turned = running lanes, so you play zone and they can buy time and pick apart a zone eventually.

so pass pressure is king with that problem we had..

as you know i hate generalizations and bullshit cliche fan speak.. if we failed lets see where, why and how and realistically address it. real data is key to reality and productive thoughts and conversations.

EPA does the things you are asking for. It's situationally based.

And the Bengals were historically bad at covering TEs in 2025.

6 hours ago, sparky151 said:

EPA does the things you are asking for. It's situationally based.

And the Bengals were historically bad at covering TEs in 2025.

DVOA does the same thing down to individual players. You could then assign weights by scheme and responsibility if known. Knowing those weights would also be a key to what players are the most important positions.

i.e. A team that plays a lot of cover 3 is going to value LB's more, and a team that plays Cover 2 is going to place a higher value on Safety. Man to man teams are going to place a higher value on CB's.

DVOA is situationally weighted. At it's heart, did the player "win" his assignment, "tie", or "lose". (Did he keep contain, block his defender, etc...) Think of this as a matrix of 2 points for a win, 1 for a tie, 0 for a loss. Given 100 plays, an average player will score 100 points. A player who scores 110 points has a 10% value over average.

So, with this chart... The Bengals gave up 46.8% more "wins" for the offense on passing plays.

Argue with how the formula is applied, but it's a solid statistical model. (As a guy who does similar work)

EDIT: Actually... DVOA has expanded a lot in the last 20 years. So that's overly simplified. By a lot!

https://ftnfantasy.com/nfl/dvoa-explainer

Edited Sunday at 05:21 AM4 days by LostInDaJungle

9 hours ago, GoBengals said:

we werent often getting diced up on tempo to TE's play after play..

That's really not true. Especially early in the season, Golden was calling a lot of cover 3 and that seam between the Mike and Safety got GASHED. Over and over again. It's a big reason why guys like Isaiah Likely had 95 yards against us in one game with a long of 43.

"Through 13 games this year, the Bengals have allowed 98 catches, 1,267 yards and 15 touchdowns against opposing tight ends."

That's 13 yards a catch. Those aren't just dump offs.

https://www.msn.com/en-us/sports/nfl/bengals-problems-against-tight-ends-are-the-worst-the-nfl-has-ever-seen/ar-AA1RWK1l

On 1/10/2026 at 10:35 PM, LostInDaJungle said:

That's really not true. Especially early in the season, Golden was calling a lot of cover 3 and that seam between the Mike and Safety got GASHED. Over and over again. It's a big reason why guys like Isaiah Likely had 95 yards against us in one game with a long of 43.

"Through 13 games this year, the Bengals have allowed 98 catches, 1,267 yards and 15 touchdowns against opposing tight ends."

That's 13 yards a catch. Those aren't just dump offs.

https://www.msn.com/en-us/sports/nfl/bengals-problems-against-tight-ends-are-the-worst-the-nfl-has-ever-seen/ar-AA1RWK1l

i think you took my comments the completel opposite of what i meant. im saying fast easy dumpoffs to TE's ARENT what we got hit with.. its Lamar jackson scrambles for 7 seconds and hits likely for 40yards and a TD after the secondary tried to cover for a long time.

On 1/10/2026 at 3:17 PM, sparky151 said:

EPA does the things you are asking for. It's situationally based.

And the Bengals were historically bad at covering TEs in 2025.

it sort of does, but not really.. unless i havent looked into it deep enough.. just comparing things to the league average isnt exactly the issue, thats still a final result vs other data, the final result isnt what i care about its what was the process before the final result... if the bengals gave up 300 field goals, and the average was 200 field goals... we suck against the average... but if the defense was put on the filed after a offensive turnover on thier own 10 yardline 285 times and we gave up 300 field goals and not TD's the defense was phenomenal, they didnt budge at all, the only better result is getting a turnover..

like burrows redzone pick 6's, just a pick there and the other team having 90 yards to go is far different than the pick 6 where they get points that magically go against the defense PPG, and lets say the pick got run back to the 5 yardline and the defense gave up a FG not a TD, that would be tremendous defensive play, but they get knocked for giving up points that were essentially guaranteed at that point no fault of their own..

this would be cool to have for offenses as well, showing the different in drive starting points and amount of drivers per game between like Bo Nix with the best defense ever and the bengals first half of the season with the worst defense ever... if that makes sense..

sort of like when two basketball players face off, but one milks like 18 free throws while the other guy is out of the game, and th statline is like "well he scored 35 points of the other player so that player blows"

but he wasnt even in the game when dud scored those point... the nba has advanced stats that are simply and clean the nfl really doesnt given so many factors and players involved. like if we had a phenominal pressure rate and DJ was a top corner, it could be he isnt great, but the pressure is making bad throws towards his guy... obviously its th eopposite with DJ, the pass rush is bad and he was great. so thats tremendous.

On 1/10/2026 at 9:50 PM, LostInDaJungle said:

DVOA does the same thing down to individual players. You could then assign weights by scheme and responsibility if known. Knowing those weights would also be a key to what players are the most important positions.

i.e. A team that plays a lot of cover 3 is going to value LB's more, and a team that plays Cover 2 is going to place a higher value on Safety. Man to man teams are going to place a higher value on CB's.

DVOA is situationally weighted. At it's heart, did the player "win" his assignment, "tie", or "lose". (Did he keep contain, block his defender, etc...) Think of this as a matrix of 2 points for a win, 1 for a tie, 0 for a loss. Given 100 plays, an average player will score 100 points. A player who scores 110 points has a 10% value over average.

So, with this chart... The Bengals gave up 46.8% more "wins" for the offense on passing plays.

Argue with how the formula is applied, but it's a solid statistical model. (As a guy who does similar work)

EDIT: Actually... DVOA has expanded a lot in the last 20 years. So that's overly simplified. By a lot!

https://ftnfantasy.com/nfl/dvoa-explainer

it does a lot but it just doesnt factor outside factors to any of the influences on those performances...

another example is if the opposing offense starts every drive on our 1 yardline, and score everytime after 4 plays.. the defense and DVOA al lsuck, but they are giving up 1 yard per drive. and .25 yards per play... so they both are terrible and great on paper and against DVOA.

1 hour ago, GoBengals said:

it does a lot but it just doesnt factor outside factors to any of the influences on those performances...

another example is if the opposing offense starts every drive on our 1 yardline, and score everytime after 4 plays.. the defense and DVOA al lsuck, but they are giving up 1 yard per drive. and .25 yards per play... so they both are terrible and great on paper and against DVOA.

First of all - I think you misunderstand DVOA. If you failed to get 1 yard 3 times, that would be 3 successes for the defense vs. every 1 success for the offense. That is then contrasted against a statistical mean defense in similar situations. So... Roundish numbers here... (If) The average is 1 in 3, 66% - and you're 1 in 4, 75% success. You're a +9% DVOA. Statistically, you would give up 9% less points than the average defense. Plus or minus the adjustments they make to "factor outside factors" as you mention which would be the area where the statistical model could become controversial.

In fact they themselves use the same example of "not all three yard runs are created equal" in their spiel for it's existence.

DVOA measures not just yardage, but yardage towards a first down: Five yards on third-and-4 are worth more than five yards on first-and-10 and much more than five yards on third-and-12. Performance is also adjusted for the quality of the opponent. DVOA is a percentage, so a team with a DVOA of 10.0% is 10% better than the average team, and a quarterback with a DVOA of -20.0% is 20% worse than the average quarterback.

Several factors can differentiate one three-yard run from another. What is the down and distance? Is it third-and-2, or second-and-15? Where on the field is the ball? Does the player get only three yards because he hits the goal line and scores? Is the player’s team up by two touchdowns in the fourth quarter and thus running out the clock, or down by two touchdowns and thus facing a defense that is playing purely against the pass? Is the running back playing against the porous defense of the Bears, or the stalwart defense of the 49ers?

Again - You can argue with the model weights, but this is a solid statistical model. You can say team X shouldn't get so many bonus points for it being cold, or that the model doesn't give enough weight to the tightness of the QB's jock strap. But statistical models like this define how AI works. Basic data science. We weight "correct responses" in an LLM just like we weight "correct plays" in a DVOA model. PFF scores are the same thing with different model weights.

The reason all of these models operate the same and get different results is because they assign different weights to different aspects of the game. PFF doesn't take into account if you're playing the Chiefs or 49ers, DVOA does. MANY people have their own weights, their own grading methodologies.... So, knock yourself out making a new one. The statistical methodology is sound.

There are plenty of "Lies and damned statistics" jokes - declaring a data model invalid because you don't like it's output is a tale older than time. But I guarantee you there's Millions if you design the one that lets Bob Kraft crack the draft. Design the model that gives a gambler an edge and you'll be rich. Figure out what aspects of the game really matter so that you have a model that correctly tells you the super bowl winner... That's the dream.

As a side note because I am a statistics nerd - your hypothetical defense would always fail on 4th down making them LESS valuable than an average defense which would succeed occasionally. That would be a failure of using DVOA - They really shouldn't be +9%. But that's also how statistics work. It's a range of possibilities, not a prediction.

I don't think there's a serious statistical model out there that will make the 2025 Bengals defense not straight-up certified unwashed dookie ass. No matter which way you turn it or how many decimal points you care to factor. You don't need algebra to confirm it.

7 hours ago, T-Dub said:

I don't think there's a serious statistical model out there that will make the 2025 Bengals defense not straight-up certified unwashed dookie ass. No matter which way you turn it or how many decimal points you care to factor. You don't need algebra to confirm it.

It's kind of like the debate in "Major League" where the 2 guys in the stands are debating "Too High" and trajectory of a home run when the third guy says "What's the difference? It's gone."

You can look at the stats 10 different ways to Sunday, but at the end of the day, our defense sucked in 2025.

13 minutes ago, Shebengal said:

It's kind of like the debate in "Major League" where the 2 guys in the stands are debating "Too High" and trajectory of a home run when the third guy says "What's the difference? It's gone."

You can look at the stats 10 different ways to Sunday, but at the end of the day, our defense sucked in 2025.

Right...

They have a major problem at NT.

Slaton isn't double teamed...no need to.

Lets another olineman freeup to assist on another dlineman...like our DEs

They need to address that issue.

21 minutes ago, claptonrocks said:

Right...

They have a major problem at NT.

Slaton isn't double teamed...no need to.

Lets another olineman freeup to assist on another dlineman...like our DEs

They need to address that issue.

Yep, and that lets Guards fire out and take out the LB's, so the LB's look bad. And then the LB's start trying to guess plays so that they have SOME chance and wind up in the wrong gap and PFF says they suck. They're over pursuing. Yadda, yadda. That's why you still have to watch film.

Building models for real world outcomes is hard. And most people have a flawed understanding of statistics anyway. There was a video game that had to adjust the percentages players saw because people got too frustrated with how often a "95% chance to hit" missed.

Saying this model didn't account for this statistical outlier... Child please. It doesn't account for the not at all real world scenario of a team constantly getting the ball on the one yard line. Boo-fuckin'-hoo. The model that DOES account for that is probably shit when it's first and ten on the 30 - a common scenario. You're starting to get into anecdotal scenarios. Next you'll be saying that the model doesn't account for Bane taking over the stadium with his henchmen. What about passes on 3rd an 11 during a solar eclipse? It's going to be all dark and DVOA will punish Chase for dropping the ball when it shouldn't!

Saying a model is flawed - it's like pointing out that water is too wet and you'd like some water, but without the wetness. I'd like to be able to go in and pick out a fish and not get my socks wet. Models are not prediction engines. They are deliberately used to model non-deterministic outcomes. When you want to model deterministic outcomes you use an algorhythm, not a statistical model. "Not 100% accurate" is a feature, not a bug. The PFF score tells you Logan Wilson sucked last week. It's an indicator, not a prediction, of how he will perform in future weeks.

If anything, we want our model influenced LESS by outliers. Many models choose to chop off the top and bottom (anomalous) X% - we don't care that one time you went 12 under when you usually go 10 over. You're not an amazing deep passer because a 70 yard hail mary got caught by your guy after it bounced off of two defender's hands, and you're not a terrible deep passer because the same hail mary with 1 second on the clock was intercepted after it bounced off of your guys hands.

"He played like an all-pro this week even though PFF has him as the worst linebacker in the league" is a comment made by an idiot. It's like saying "If there's Global Warming, why is it cold today?". It's an argument made by people too stupid to understand why it's dumb thinking they're being smart. Myles Murphy has one good play and people in here lose their minds thinking they've finally proven that PFF is a lie engineered by the Rothschilds and the lizard people living under the Denver airport. "Ah hah!"

It's the old problem as assuming that a coin flip is 50/50, the last flip was heads, so statistically, the next flip has to be tails. Then someone flips heads 10 times in a row and acts like the math is busted. "It can't be 50/50 because I got heads 10 times in a row". Once you get past a 6 sided dice, humans lose the ability to accurately understand probability. (This is why 6 sided dice are popular in gambling!) Even then... He just rolled 7 three times in a row, the dice must be loaded. He's cheating! (you have a 20% chance of rolling a 7 on each throw - 1 in 5)

A few weeks ago I talked about the effect of just one extra hit in the draft. How drastic changing the house advantage by even half a percent can be. If you had a model that gave a bettor an extra half a percent edge - you're card counting. So there are many people working on this. Even the betting companies that set the line want this tech. Sabermetrics, Billy Bean... Figuring this out would make you the greatest GM in the league - and pay Millions each year. I assume that many of the "hot" GM's and teams have their own model and weights for internal use.

I have heard people say that this model favors that team, or that XYZ site loves Steeler fans... There's too much money on the line for that to be feasible. Not to mention too many small tweaks to a large and complicated model. The foundation exists and is easy enough to understand... If you think you have the ability to curate and weight the data in a meaningful way, then go for it. It'll make you a rich and powerful man.

16 hours ago, GoBengals said:

i think you took my comments the completel opposite of what i meant. im saying fast easy dumpoffs to TE's ARENT what we got hit with.. its Lamar jackson scrambles for 7 seconds and hits likely for 40yards and a TD after the secondary tried to cover for a long time.

it sort of does, but not really.. unless i havent looked into it deep enough.. just comparing things to the league average isnt exactly the issue, thats still a final result vs other data, the final result isnt what i care about its what was the process before the final result... if the bengals gave up 300 field goals, and the average was 200 field goals... we suck against the average... but if the defense was put on the filed after a offensive turnover on thier own 10 yardline 285 times and we gave up 300 field goals and not TD's the defense was phenomenal, they didnt budge at all, the only better result is getting a turnover..

like burrows redzone pick 6's, just a pick there and the other team having 90 yards to go is far different than the pick 6 where they get points that magically go against the defense PPG, and lets say the pick got run back to the 5 yardline and the defense gave up a FG not a TD, that would be tremendous defensive play, but they get knocked for giving up points that were essentially guaranteed at that point no fault of their own..

this would be cool to have for offenses as well, showing the different in drive starting points and amount of drivers per game between like Bo Nix with the best defense ever and the bengals first half of the season with the worst defense ever... if that makes sense..

sort of like when two basketball players face off, but one milks like 18 free throws while the other guy is out of the game, and th statline is like "well he scored 35 points of the other player so that player blows"

but he wasnt even in the game when dud scored those point... the nba has advanced stats that are simply and clean the nfl really doesnt given so many factors and players involved. like if we had a phenominal pressure rate and DJ was a top corner, it could be he isnt great, but the pressure is making bad throws towards his guy... obviously its th eopposite with DJ, the pass rush is bad and he was great. so thats tremendous.

it does a lot but it just doesnt factor outside factors to any of the influences on those performances...

another example is if the opposing offense starts every drive on our 1 yardline, and score everytime after 4 plays.. the defense and DVOA al lsuck, but they are giving up 1 yard per drive. and .25 yards per play... so they both are terrible and great on paper and against DVOA.

EPA (expected points added) is a plus/minus number which can be judged per game, per drive, or per play. Suppose a team has 1st and goal from the 1 yard line. There's a high probability they'll score a touchdown and kick an extra point. But suppose the QB throws a pick in the endzone. That's a negative result for him of almost 7 points he cost his team (since the touchdown and xp aren't guaranteed). If the endzone pick is returned for a pick six and xp, then it's almost negative 14 points for him.

On 1/10/2026 at 3:07 PM, GoBengals said:

we dont have an issue covering TE's nearly as much as we had a pressure on the QB problem, we werent often getting diced up on tempo to TE's play after play.. we werent getting to the QB, they bought time, and the TE is both often covered by the weakest cover guy or in zone time lets you find a hole in zone, and WR are priority, LB safety etc often end up covering TE's.

On 1/13/2026 at 12:49 AM, GoBengals said:

im saying fast easy dumpoffs to TE's ARENT what we got hit with

No, we got hit with guys running full on routes up the seam in Cover 3 between the Safety and Mike. 3 step drop, fire as soon as the back foot hits. It had nothing to do with the Pass Rush.

I mean, if I was an offensive Coordinator looking at the Bengals D on paper... Of course I'm going to stress the two rookie LB's. Duh. I'm going to use my TE's and Slot WR to hit those seams, hi-lo them... Let's check those ball skills rook. I mean, hope springs eternal and all of that... But did you really think that two Rookies were going to suddenly going to be shutting down Mark Andrews? We lost Hubbard, got TJ Slayton, and let go our two veteran "star(ish)" LB's. Brought in a rookie D-Coord to replace "The Wizard". And you thought we'd improve?

The answer is that we had a HUGE problem covering TE's. AND we had a terrible pass rush. That's how you get a historically bad performance - the worst of the modern era. There's no way to candy coat that. No level of the defense was adequate. Carter and Knight weren't at the bottom of the PFF rankings due to an illuminati conspiracy. They really did play badly. They really did give up a lot of yards and points to bad teams.

I mean, no offense, but you really seem to decide what you'd like to believe and then work backwards. And you've been scrambling for a while now to try and justify our LB situation. From Logan Wilson going from team leader to washed up bum in 8 games, the hate on Pratt...

So - There's good news and bad news. The Rookie LB's aren't terrible. But neither were Pratt or Wilson. Both pairings suffered from having a D-Line that can't command a double team and can't contain the run. The problem wasn't the linebackers.

It's very much defense 101. If you can't trust BJ Hill to be stout against the run, you hesitate at the snap and now the TE is gashing you up the seam. If Slayton can't command a double team, you have a 300 pound guard blocking a 230 pound LB in the run game - advantage offense. The RB has clear lanes to come at you full speed with wiggle room - so you "miss" tackles. The same factors that made Wilson and Pratt look a step slow... They make Carter and Knight look like dog crap too.

I mean, it says something that LB play got MUCH better when DE depth got thin and Golden stopped using Murphy/Ossai/Stewart as an undersized three tech. Between that and a general lack of effort...

IMHO - McKinley Jackson was mouthing off in week 6 because he could see, just like I could, that we should be getting a real NT more snaps.

Check my post history - I'm big on Knight. Carter is OK in my book too. But you wouldn't start a second round QB in Game 1 and expect a great offensive performance. You wouldn't do it behind a crappy O-Line. And I don't think it's "mean" to say "Damn he had a rough day behind that crappy O-Line." You might even be inclined to say we should hold on to an Andy Dalton and let the rookie second rounder grow into the position... I mean, if you weren't all up in arms because Germaine Pratt makes more than you think he should. IMHO - you hold on to that depth. Do you know what we used the money we saved on Pratt and Wilson to do?? Nothing.

In many ways Carter and Knight are just as guilty of giving up 13 YPC to TE's as Joe Burrow is for taking 9 sacks in a playoff game. You can argue that Burrow held on to the ball to long, should have checked to a different play... You can just blame the line. There is no "right" answer. What you can't do is pretend that the line didn't give up 9 sacks and plan to do nothing about it in the future. Or assume that the problem will just fix itself. Figuring out an argument for why Knight is actually a really good player doesn't give the Jets less points or stop Cleveland from scoring in the last two minutes.

6 minutes ago, LostInDaJungle said:

The answer is that we had a HUGE problem covering TE's.

This isn't a this year problem. This has been an issue for years. Add in the tackling issues and it's been a recipe for disaster, especially when they're trying to tackle TEs like Washington from Pittsburgh who is the size of a semi truck.

What this defense needs is a true leader, someone who isn't afraid to get physical, someone who will tackle ferociously but legally, an enforcer type.

6 minutes ago, Shebengal said:

What this defense needs is a true leader, someone who isn't afraid to get physical, someone who will tackle ferociously but legally, an enforcer type.

Guys like that don't smile enough for Ol' Musty

4 hours ago, Shebengal said:

This isn't a this year problem. This has been an issue for years. Add in the tackling issues and it's been a recipe for disaster, especially when they're trying to tackle TEs like Washington from Pittsburgh who is the size of a semi truck.

What this defense needs is a true leader, someone who isn't afraid to get physical, someone who will tackle ferociously but legally, an enforcer type.

Agree but at what position?

The hardest hitters we have are who?

It's tackling that's preimminent..

Who in the hell teaches this tackling system ?