Notable Stats and Observations

[Danger]

57 minutes ago, skywlker32 said:

So the point was to combine two trash evaluation tools and watch chaos ensue?

12:7 doesn't equate to MVP lol.

[skywlker32]

2 minutes ago, Danger said:

12:7 doesn't equate to MVP lol.

?

[Danger]

Just now, skywlker32 said:

?

sorry meant to quote you and that post you quoted with the tweet suggesting Dak was an "MVP candidate"

[Hunter2_1]

Going in to WEEK 9

 

MASON RUDOLPH TOOK HIS SHOTS ON MONDAY NIGHT

In Weeks 2 to 5, Mason Rudolph’s average pass attempt traveled just 6.7 yards — 29th among 34 quarterbacks with at least 50 attempts in that span. In his return to action against the Miami Dolphins, though, Rudolph let loose. He ranked first in average depth of target in Week 8, averaging 13.1 yards through the air on attempts.

 

GARDNER MINSHEW ON PLAY-ACTION? UNBEATABLE.

Gardner Minshew entered Week 8 with the fourth-best passer rating on play-action. By the end of Sunday and Monday’s games, he rose up the rankings to No. 1. Minshew posted a perfect 158.3 mark against the Jets, completing all five of his play-action passes for 117 yards and a touchdown.

 

FEEDING DAVID MONTGOMERY SEEMS TO BE A GOOD IDEA

Mitchell Trubisky simply isn’t getting it done for the Bears. Montgomery, on the other hand, is doing his part. He had 10 first-down carries, six missed tackles forced and 64 yards after contact in Week 8 — all of which ranked top-10 among qualifying running backs.

 

BENGALS RECEIVERS ARE GETTING TARGETS 

Andy Dalton is out under center in Cincinnati, and likely for good reason. No Bengals wideout with at least 20 targets has caught more than 70% of balls thrown their way, and Auden Tate and John Ross (currently on IR) rank 85th and 86th, respectively, out of 90 players in completion rate when targeted.

 

NICK BOSA IS DRAWING ALL THE HEADLINES, BUT IS THERE ANYTHING BIG BRO CAN’T DO?

The Bosa brothers are something special, and Joey’s skillset was on full display Sunday. While his eight quarterback pressures were far from surprising, Bosa’s six stops proved to be a new career-high for the fourth-year man out of Ohio State.

 

PRESSURE DID NOT FAZE RUSSELL WILSON IN WEEK 8

Five quarterbacks have posted a perfect passer rating under pressure this season, with three of those efforts coming in Week 1. Russell Wilson added his name to the list Sunday, completing 6-of-7 pressured attempts for 89 yards and a touchdown.

 

KEVIN ZEITLER KEPT DANIEL JONES CLEAN AGAINST THE LIONS

For the second time this season, Kevin Zeitler didn’t allow a single quarterback pressure. And he did it largely against Da’Shawn Hand — who played in his first game of the season after finishing his rookie campaign with a strong 87.4 overall grade.

Edited October 31, 2019 by Hunter2_1

[Spartacus]

1 hour ago, ramssuperbowl99 said:

There are ways to see if a statistic is value added or not without knowing the exact formula. For example stuff that's a thousand differential equations that you have to solve numerically with a computer, nobody is going to be able to simplify that to a standard algebraic formula, but those models are still very valuable.

You just take a bunch of previous data, split it into a test sample (e.g. future performance) and validation sample (e.g. past performance), then you see how well that stat does. So you could do this with QBR, total QBR, QB rating, etc. etc. The more predictive the stat is using previous data, the more likely it will predict better using future data.

That's how most baseball stats work. So FIP predicted better than ERA, xFIP predicted better than FIP, and dRA predicted better than xFIP. Here's BP doing that exercise when they introduced DRA:

https://www.baseballprospectus.com/news/article/26195/prospectus-feature-introducing-deserved-run-average-draand-all-its-friends/ 

Any statistical value worth anything will be left in the open to be analyzed for weakness. The fact that it has been kept "Secret" tells me that its most likely extremely flawed in some fashion. 

[Spartacus]

50 minutes ago, Danger said:

QBR is closer to the Dunk Contest than it is a legit statistic. I don't remember who it was, but multiple posters here before have said that ESPN changed someone's QBR days later after a game because it seemed so poorly graded compared to the opposing QB's performance.

It really is irritating when I see people spout QBR. I don't know why but the stat just bothers me. 

[ramssuperbowl99]

11 minutes ago, Spartacus said:

Any statistical value worth anything will be left in the open to be analyzed for weakness. The fact that it has been kept "Secret" tells me that its most likely extremely flawed in some fashion. 

Again, not true. Plenty of proprietary stuff is kept secret. All statistics and models are "wrong", by definition. They're bite-sized oversimplifications of reality. The question that matters is whether or not they work, and that can be answered without knowing the exact math.

[Spartacus]

8 minutes ago, ramssuperbowl99 said:

Again, not true. Plenty of proprietary stuff is kept secret. All statistics and models are "wrong", by definition. They're bite-sized oversimplifications of reality. The question that matters is whether or not they work, and that can be answered without knowing the exact math.

I mean even when we use your opinion QBR does not stand up as being an accurate representation of anything:

Aaron Rodgers: 369 yards, 4 TDs, 1 int, and his QBR was 24.9

Brandon Weeden: 232 yards, 0 TDs, 1 int, and his QBR was 43.9

 

Aaron Rodgers: 31-36, 396 Yards, 2 TDS, 0 int, and his QBR was 82.1

Tim Tebow: 4-10, 73 Yards, 1 TD, 0 int, and his QB was 83.2 

 

 

 

These are just a couple examples but you can literally see them every year. QBRs goal is to judge the success of an individual QB performance but its so piss poor at it then what the hell is the point? Its a worthless stat parroted like it actually means something when it obviously is deeply flawed. 

 

[ramssuperbowl99]

10 minutes ago, Spartacus said:

I mean even when we use your opinion QBR does not stand up as being an accurate representation of anything:

Aaron Rodgers: 369 yards, 4 TDs, 1 int, and his QBR was 24.9

Brandon Weeden: 232 yards, 0 TDs, 1 int, and his QBR was 43.9

 

Aaron Rodgers: 31-36, 396 Yards, 2 TDS, 0 int, and his QBR was 82.1

Tim Tebow: 4-10, 73 Yards, 1 TD, 0 int, and his QB was 83.2 

 

 

 

These are just a couple examples but you can literally see them every year. QBRs goal is to judge the success of an individual QB performance but its so piss poor at it then what the hell is the point? Its a worthless stat parroted like it actually means something when it obviously is deeply flawed. 

 

This isn't my opinion. It's how models are validated. Saying "we don't have the formula" is not a viable criticism. Pointing out 4 points that don't fit a dataset is not a viable criticism - that's like saying a climate model is bad because it was unseasonably warm yesterday.

 

I don't know much about QBR, so I'm not even sure if it's a good stat or not. But I do know you're going nowhere fast trying to criticize it the way you have.

[skywlker32]

9 minutes ago, ramssuperbowl99 said:

This isn't my opinion. It's how models are validated. Saying "we don't have the formula" is not a viable criticism. Pointing out 4 points that don't fit a dataset is not a viable criticism - that's like saying a climate model is bad because it was unseasonably warm yesterday.

 

I don't know much about QBR, so I'm not even sure if it's a good stat or not. But I do know you're going nowhere fast trying to criticize it the way you have.

Part of the problem is that QBR seems to take rushing into account WAY too heavily.

As for the importance of the model correlating to a win, I can make a model that "predicts" a win by incorporating offensive points scored and the actual outcome of the win, but no one would know that if I keep it as a proprietary model. For all we know, they make up a number based on the "eye test" since they supposedly changed the QBR stat without telling anyone.

This gives a good summary of the issues with QBR:https://roundballdaily.com/2016/01/13/espn-got-called-out-for-their-terrible-qbr-rating-and-then-changed-it/

[ramssuperbowl99]

3 minutes ago, skywlker32 said:

Part of the problem is that QBR seems to take rushing into account WAY too heavily.

As for the importance of the model correlating to a win, I can make a model that "predicts" a win by incorporating offensive points scored and the actual outcome of the win, but no one would know that if I keep it as a proprietary model. For all we know, they make up a number based on the "eye test" since they supposedly changed the QBR stat without telling anyone.

This gives a good summary of the issues with QBR:https://roundballdaily.com/2016/01/13/espn-got-called-out-for-their-terrible-qbr-rating-and-then-changed-it/

This still could be the most predictive statistic. What you put into a model only matters in as much as you want to improve the model performance.

If, hypothetically, you had 2 draft models that you were comparing. One was some complicated algorithm of combine data, and college performance, and age and tons of other stuff, while the other was Bill Belichick's personal rankings that he just made up. If you took 20 years, and BB is outperforming the algorithm, go with BB. It works better.

That's the whole goal. To find something that works for the population of data. How you get there, whether is some algebraic formula, or differential equation you need to solve numerically with a computer, or a homeless guy giving you draft advice, it does not matter. Prove what works best, and use that.

[skywlker32]

4 minutes ago, ramssuperbowl99 said:

This still could be the most predictive statistic. What you put into a model only matters in as much as you want to improve the model performance.

If, hypothetically, you had 2 draft models that you were comparing. One was some complicated algorithm of combine data, and college performance, and age and tons of other stuff, while the other was Bill Belichick's personal rankings that he just made up. If you took 20 years, and BB is outperforming the algorithm, go with BB. It works better.

That's the whole goal. To find something that works for the population of data. How you get there, whether is some algebraic formula, or differential equation you need to solve numerically with a computer, or a homeless guy giving you draft advice, it does not matter. Prove what works best, and use that.

That's not a statistic though, that's an opinion. Statistics should be facts. In the site I posted, it explains that Charlie Batch's 3TD 2INT game was rated as the best game EVER. There is no model that should show that objectively. That isn't a statistic or model, that is just straight up non-sense.

[El Ramster]

The top 3 QBs by redzone passer rating this year are Case Keenum, Jared Goff and Teddy Bridgewater. 

 

 

jared Goff nation stand the hell up. 

 

 

[ramssuperbowl99]

1 minute ago, skywlker32 said:

That's not a statistic though, that's an opinion. Statistics should be facts.

The best, most accurate pain models are based on aggregating people's 1-10 scores for how much pain they're in. There are 0 "facts" in there, just "opinions". But your post surgery care is going to be the best it can be when you're coming out if the nurse asks you how you feel, scale of 1-10. So, again, this is just wrong.

[Chargers]

On 10/28/2019 at 10:52 AM, animaltested said:

The Seahawks are 6-2, 4-0 on Road, three of those wins on the East Coast at 10am. But the Seahawks fanbase, on the whole is teetering on the edge of a chasm of nihilism (including myself). 

This season has been... very weird in many ways. 

Sounds like spoiled fanbase. You're 6-2 and have Wilson playing at a MVP level. What more can you ask for at this point?