Clary, why is he still here???

[Yubaking]

Not the 'they' you quote, but rather some of the posters involved in this debate. Seems that some PFF stats have meaning and others don't based on personal biases towards the players involved. Yuba uses PFF as a basis to bash Clary and ignores it in bashing Freeney.

SD17, in the long Freeney thread, what I said (and I went back and looked) was that Freeney's pressures meant almost nothing because at the time even with his pressures, we were still surrendering a 70% completion percentage by opposing QBs despite then getting pressures on 43.8% of passing snaps. And we had no turnovers period much less any because of generating QB pressure.

Steve has since made what I always felt was a good point that Freeney's pressures did have more value, but we couldn't cash in on that value because nobody else was there to sack the QB when Freeney got pressure. He was optimistic that Freeney could combine with Ingram to do some real damage. I think that argument makes sense and it gives me hope that Freeney's impact will be greater for us this year than it was last year.

As for PFF, at first I rejected their analysis pretty much altogether and so when you say that I was against it and didn't accept it, you are right. Some of the results did not seem right in terms of what I saw and I had the same problem with PFF that Steve has complained about--that their precise scoring rules have not been published. I have that same issue with ESPN's Total QBR system and I do think that that problem does count against the credibility of PFF's specific grades.

However, it has occurred to me since that time that in cases where the PFF grades are more extreme (heavily positive or negative), the PFF folks would have had to have really screwed up to have missed whether or not the player was positive or negative at all. Giving them the benefit of that small doubt, I do think it is fair to say that in extreme cases that it is at least more likely than not that PFF's general conclusion whether or not the player really was positive or negative (good or bad) is correct. Is it conclusive proof? No, but it is enough to I think to assert a general conclusion about the player in question in those limited instances.

The limited exception does not apply to Freeney. He was slightly positive for the season, but trending negatively after the first game (per OIP, he started at +6.1, but as information provided by Beerman showed us, he finished positively, but less than +6.1 (something like plus 4 and change)).

I think the limited exception that I am employing with respect to PFF's grades for the purpose of making only a very general qualitative statement about the player's performance (either good or bad) is really just common sense.

So, the general rule I am using is that PFF grades are not very useful, but the corollary to that rule is that the more extreme grades probably are slightly more reliable in terms of formulating a general conclusion about a player's play and therefore do have at least some value. And under that system of viewing things, I think the scores of Dunlap and Clary are more reliable than the score of Freeney, and frankly, the scores of lots of other players.

Because I have modified my position, I think your claim of inconsistency against me is somewhat fair.

[blueman]

OIP, I have said this countless times, but what did YOU think?

That question would make sense if oip came here to talk football. Unfortunately the overwhelming majority of his posts would indicate he comes here for something else. Is what it is.

[oneinchpunch]

Not really an informed. A misinformed (at least for OL), which is why I am surprised you don't like them more.

I think their analysis of Individual OL play as well
As DL play is better than most other sites.

It was expected that Fluker would struggle some that game but Clary did as well as he did for most of the season. I like Clary and hope he has a chance to stay and improve this season but he's had back to back season where he's been subpar.

[RobH]

If Clary is still the starter at the beginning of this next season, then I'm going to have to conclude that PFF and all the other sites are incorrect. TT has had plenty of chances to replace him, and hasn't done it yet. He may still do so through the draft, but I really don't think so. They may still replace him with Troutman or some other OG still on the team or with an FA we still haven't signed. I really don't think that will happen, though. And if Clary is still the starting RG, then it must mean that Coach D, whom we've been raving about as being a darn good coach, must have a good reason not to. So I guess it remains to be seen if Clary is so bad that he is demoted from his starting job, but if he isn't, then I would expect some people to admit that they don't know OL play as well as they think. That is, unless they think that they know more than Coach D and the other coaches on our team.

[Panama]

Watt's on second.

I don't know.

[TTK]

Riddle me this. If Clary is as dogshit as his grades say he is, why the hell wasn't he cut long ago to upgrade, especially at his ridiculous salary? They're basically saying he's only better than about 10% of the guards out there. If he's that bad, they shouldn't want him at all even as a backup at the vet minimum. There are guys on the streets that graded higher and would be much cheaper.

Or maybe TT, MM and Coach D know infinitely more about the intricacies about OL play over some dudes who compile stats on a website and some guys who like to post on message boards. You know, because they've studied, played, coached, scouted this for their entire career.

[thelightningwill]

E

Riddle me this. If Clary is as dogshit as his grades say he is, why the hell wasn't he cut long ago to upgrade, especially at his ridiculous salary? They're basically saying he's only better than about 10% of the guards out there. If he's that bad, they shouldn't want him at all even as a backup at the vet minimum. There are guys on the streets that graded higher and would be much cheaper.

Or maybe TT, MM and Coach D know infinitely more about the intricacies about OL play over some dudes who compile stats on a website and some guys who like to post on message boards. You know, because th
ey've studied, played, coached, scouted this for their entire career.

You obviously didn't write Money Ball.

[blueman]

Too much sanity, ttk.

[Zot]

I don't know.

...third base.

[Steve]

The fundamental problem with football and stats is that it is not as simple. Look at it this way, can you have a great run blocking OL on a team that cannot run the ball? Sure, it happens all the time. 1 OL performance really tells you nothing about how the running game. You need to know how every guy blocks. And even if everyone blocks well, you could still do badly. Say the other team brings their SS down into the box, so that there is an extra guy to block. A lot of running plays only work against certain types of defenses (7 vs 8 man fronts). And how do you account for the plays where the QB should check out because he can read that presnap, vs the plays where the secondary rotates at the snap. How do you account for players who have tougher blocking assignments then others? How about if the D shifts or stunts/blitzes in such a way that the blocking cannot account for. If you don't remove these sorts of biases from your grading, it is going to be impossible to construct a grade that accurately accounts for a players "true" contribution.

One of the things that happens in science is that the guys doing the science have to be responsible enough to use an analysis method that really fits the problem being investigated. In the US (middle latitudes) there has never been a successful statistical model for forecasting the weather. It just doesn't work. So, the techniques used focus on a more direct method. You literally break the US into a grid, find the basic equations of fluid motion, and use multiple grids at each level of the atmosphere, and calculate where weather systems are moving directly with a super computer.

My point is, that while Moneyball points to how statistics was used in baseball the same methods will not work necessarily work in football. Maybe there is a way to remove some of the biases and then do some reasonable stats, or alternatively, do some more elaborate analysis of multiple variables (multivariate) statistics.

[Mister Hoarse]

The fundamental problem with football and stats is that it is not as simple. Look at it this way, can you have a great run blocking OL on a team that cannot run the ball? Sure, it happens all the time. 1 OL performance really tells you nothing about how the running game. You need to know how every guy blocks. And even if everyone blocks well, you could still do badly. Say the other team brings their SS down into the box, so that there is an extra guy to block. A lot of running plays only work against certain types of defenses (7 vs 8 man fronts). And how do you account for the plays where the QB should check out because he can read that presnap, vs the plays where the secondary rotates at the snap. How do you account for players who have tougher blocking assignments then others? How about if the D shifts or stunts/blitzes in such a way that the blocking cannot account for. If you don't remove these sorts of biases from your grading, it is going to be impossible to construct a grade that accurately accounts for a players "true" contribution.

One of the things that happens in science is that the guys doing the science have to be responsible enough to use an analysis method that really fits the problem being investigated. In the US (middle latitudes) there has never been a successful statistical model for forecasting the weather. It just doesn't work. So, the techniques used focus on a more direct method. You literally break the US into a grid, find the basic equations of fluid motion, and use multiple grids at each level of the atmosphere, and calculate where weather systems are moving directly with a super computer.

My point is, that while Moneyball points to how statistics was used in baseball the same methods will not work necessarily work in football. Maybe there is a way to remove some of the biases and then do some reasonable stats, or alternatively, do some more elaborate analysis of multiple variables (multivariate) statistics.

I've read some books on football written by some great minds like Steve Bellichik, Pat Kirwan, etc. But you have taught me as much or more about the intricasies of the game by posting here and at the old board. I want to acknowledge that in this somewhat ridiculous thread.

[thelightningwill]

...third base.

That's what she said.