• Changing RCF's index page, please click on "Forums" to access the forums.

2019 Cleveland Browns Regular Season

Do Not Sell My Personal Information
I made a thread about this on Twitter, but in case people are curious (just humor me), the analytics models are actually uniformly picking the Browns to finish between 16th-18th in the NFL this year. I want to explain why.

(sorry for hijacking this thread, @Juice Is Loose, loved the Myles article).


1) Uncertainty about the head coach. Kitchens is a new HC, and new HCs tend to be unstable. My guess is that most models have Freddie making a negative impact (a neutral impact would be league average/team's schedule). Even more, I imagine those models have Freddie adding anywhere from half-to-one extra losses.

2) Baker Mayfield is going into his second year. This combines with the fact that his play under Hue was pretty underwhelming. Consequently, models viewed Baker as a league average QB last year, and project slight improvement. Expected points added models have Baker as a +2.5 win QB. Mahomes is a +5 and highest in the NFL.

3) Lots of roster overhaul adds even more instability. The Browns have eight new starters, five new additional rotation players, and unknown philosophies on both sides of the ball.

4) The offensive line is predicted to be significantly below average. Historically, the team only has one elite linemen and two average starters. Robinson is a big question mark and Kush is not that good.

5) The depth at a lot of positions just are not all that great. QB, WR, all offensive line positions, and all defensive line positions will be much worse in case of injury.

6) The Browns were 7-8-1 last season, 5-3 after Hue got fired, and only got one win against a playoff team. They played .500 football all year. The biggest positions they added this year are at receiver and secondary pass rusher - two positions reliant on other factors.

7) Compounding all of this is that the Browns play two games against the Ravens - who are another team with very similar problems - and the Steelers - who are actually a really reliable eight-to-ten win pick. The schedule may look easy, but it is actually pretty unpredictable.

I think this leads to two valuable pieces of information:

First, the best models we have still do not perform greatly at incorporating context. ELO models view the Browns as a slightly below .500 team, with lots of overhaul, whose QB will add about one more win than their total last season. They cannot incorporate how the players feel about Freddie, if they know his offense, how good Baker looked under Freddie, etc.

Second, with all of that said, the models do identify uncertainty with this Browns team. Incorporating new players is difficult, no matter the coach. Injuries are unpredictable and the Browns will be impacted more than most by such events. Overall, if you are risk averse, the Browns probably are not a team you want to root for.

Still, I think the models are underrating the Browns by about thirty total net points (which roughly equates to one win). My guess is the Browns go 9-7 and make the playoffs. If they avoid injuries, though, the sky is the limit.
 
There's a good chance Rene Bugner is my wife's secret pen name, because this article is a case study in nit-picking.

The Browns have been the most aggressive in changing over the offseason. I agree that those who are risk adverse wouldn't bet on the Browns... but what fun is it to predict another Patriots AFC Championship win over Ben Roethlisberger? Dare to eat a peach. Dare to have fun. Dare to invest your Sundays in the Browns.
 
I dislike how uncertainty is always viewed as BAD. Freddie being a new HC always seem to be viewed negatively rather than positively or simply neutral.

I could list 20 LEGITIMATE reasons why the Browns should be improved, vs about 5 hypotheticals (egos! new HC!) on why the Browns should regress.

The NFL is unpredictable and by no means am I guaranteeing anything, but how anyone can project the Browns to be WORSE is mind-boggling to me.
 
There's a good chance Rene Bugner is my wife's secret pen name, because this article is a case study in nit-picking.

The Browns have been the most aggressive in changing over the offseason. I agree that those who are risk adverse wouldn't bet on the Browns... but what fun is it to predict another Patriots AFC Championship win over Ben Roethlisberger? Dare to eat a peach. Dare to have fun. Dare to invest your Sundays in the Browns.
Just to be clear, I was not trying to say that the reasons are legitimate, just trying to explain why the analytic models have the Browns with a lower win total than most football evaluators.

I dislike how uncertainty is always viewed as BAD. Freddie being a new HC always seem to be viewed negatively rather than positively or simply neutral.

I could list 20 LEGITIMATE reasons why the Browns should be improved, vs about 5 hypotheticals (egos! new HC!) on why the Browns should regress.

The NFL is unpredictable and by no means am I guaranteeing anything, but how anyone can project the Browns to be WORSE is mind-boggling to me.
When I say uncertainty, I mean statistical uncertainty, and that is not necessarily a bad thing. It is just language of prediction.

So the way the models work is via regression. The regression produces a standard error, and because these variables create uncertainty, this standard error is really large. Nobody shares their results, but my guess is the Browns standard error with a 99% confidence interval ranges between six and twelve wins. It is probably the largest in the NFL, along with the Ravens.

Then the regression tries to establish the significance of every win total... because the error is so large, and because of their performance last season, the most confidence is probably at 8-8.

Again, I think the models are adding uncertainty that is not actually present because of how their formulas function. Literally all I was trying to do is explain why the 538, PFF, Football Outsiders, and ELO models have the Browns at 8-8.
 
Just to be clear, I was not trying to say that the reasons are legitimate, just trying to explain why the analytic models have the Browns with a lower win total than most football evaluators.


When I say uncertainty, I mean statistical uncertainty, and that is not necessarily a bad thing. It is just language of prediction.

So the way the models work is via regression. The regression produces a standard error, and because these variables create uncertainty, this standard error is really large. Nobody shares their results, but my guess is the Browns standard error with a 99% confidence interval ranges between six and twelve wins. It is probably the largest in the NFL, along with the Ravens.

Then the regression tries to establish the significance of every win total... because the error is so large, and because of their performance last season, the most confidence is probably at 8-8.

Again, I think the models are adding uncertainty that is not actually present because of how their formulas function. Literally all I was trying to do is explain why the 538, PFF, Football Outsiders, and ELO models have the Browns at 8-8.
Actually, this is a great article that explains some of what I am getting at...


Again, I think a big problem with analytics is that it cannot consider context. I disagree with its evaluation of the Browns. But these are reasons why they are lower on the Browns than most football watchers.

"Players who are hot late in the season have big things ahead in the next season"

This myth is coming into play now with the preseason hype around the Cleveland Browns and Baker Mayfield. Mayfield was absolutely outstanding in the second half of last season. Based on our passing DVOA ratings, his improvement from the first half of the season to the second half was the third-largest of any quarterback since 2004. Certainly, that improvement is going to carry over to this season, right?

Well, the first indication that it's not is the name of the quarterback with the greatest second-half improvement in the past 15 years: Joey Harrington for the 2005 Lions. Harrington didn't exactly set the football world on fire in 2006.

Throughout the history of Football Outsiders, we've looked at the idea that improvement or decline in the second half of a season predicts performance the following season. We always get a similar result: There doesn't seem to be a pattern where players or teams with outstanding second halves carry that momentum over to the next season. Because of Mayfield, we looked at it this offseason, limited to only first-year starters who improved in the second half of that season. Even for these players, their full-season performance was a better guide to their performance the following season than their hot second halves.

There's plenty of reason to believe Mayfield will be an excellent quarterback this season. He was very good for a rookie to begin with, and now he's adding Odell Beckham Jr. to his wide receiver corps. But that hot second half, specifically, is not a reason to believe Mayfield will be great this season.

"'Analytics' says teams should always pass"
Most of myth-busting here has been about how the analytics community views the game of football. This one is more about how football fans view the analytics community.

Yes, analytics says that NFL teams need to pass the ball more often. But it's incorrect to exaggerate and state that analytics says teams should always pass the ball.

First, there are definitely situations where running is more efficient than passing, primarily short-yardage situations. To use just last season as an example, runs converted 73% of the time on third-and-1, while passes converted just 59% of the time in those situations. In fourth-and-1 situations, those numbers were 75% for runs and 64% for passes.

Second, we can analyze the game only as it is actually played, both now and in the past. That means that in every game we analyze, there's always the threat of the run. If there was a football team that literally never ran the ball, it would change the way defenses would play against that team. That would change the probabilities of certain decisions. Most people in the analytics community understand this. We're suggesting that teams pass the ball more, not that they eschew the run completely.
 

Rubber Rim Job Podcast Video

Episode 3-13: "Backup Bash Brothers"

Rubber Rim Job Podcast Spotify

Episode 3:11: "Clipping Bucks."
Top