Separate names with a comma.
Discussion in 'Miami Dolphins Forum' started by hitman8, Aug 28, 2017.
I like the Miami Dolphins and he plays for them
Seriously though, just curious. If he has 10 DEs outperforming him, in PFF's eyes, and all of them are playing roughly the same level of competition, that's interesting. Perhaps he is coming along more slowly than them. If he has 2 DEs outperforming him, with similar level of competition, then he's right where his draft position would suggest.
So.. with all this pff stuff around here I decided to try and (partially) answer the question of whether there's any objective evidence that pff ratings aren't that good.
First place I looked was their prediction of win totals for the 2016 season:
Compare that to archived odds from Vegas for 2016:
Correlation between pff predicted win totals and actual win totals is 0.4489 while Vegas beat them at 0.5161. One would have to do this for a bunch of years to see what the trend is but at least one can say pff isn't somehow using information better than those that aim to make money off of their predictions!
Perhaps more revealing is comparing two different ratings to win%: pff ratings for QB's and passer rating. In both cases QB's are being rated after the fact so that part's comparable. For pff we have the ratings from near the end of 2016:
And passer ratings for those QB's that qualified as starting QB's on pro-football-reference (sort by passer rating to filter):
OK, first of all pff rated 32 QB's and there are only 30 QB's that qualified as "starters" on PFR. Furthermore, Keenum is on PFR but not ranked in pff in that link. So we're comparing 29 QB's here and their win% based on games started.
Correlation between passer rating and win% for those 29 QB's was 0.5382 while the correlation between pff ratings and win% was a dismal 0.3627!
OK.. THAT is a big difference. This means that the simple passer rating formula that only looks at Y/A, COMP%, TD% and INT% does better than human graders looking at everything related to a QB with regards to how it influences win%.
Where precisely is the difference? Compare the two:
and something interesting can be seen. PFF graders are either: 1) overweighting the value of average QB's, or 2) underweighting the value of the worst QB's, all relative to what passer rating does, because that's what it would take to better fit a straight line between the two sets of ratings.
So at least for predicting overall ability of a team and for estimating the influence on win% of starting QB's, there are better sources than pff. Importantly, pff doesn't do a good job of evaluating a QB's influence on win% when compared to a simple stat like passer rating.
No, I really didn't read your post.
Congrats to this bust for his 1st career sack, hopefully he won't get many more.
Yeah! If he plays well, what will I complain about?!?!?!
I am very happy that he played well today. He finally showed up and had a great game.
But up until this point he had done absolutely nothin. He does well against backups but gets neutralized when he plays a decent tackle. It was the same thing in his college career all of his production came against scrubs. The titans tackle lewan was playing hurt and the backup had to step in. Andre branch also had a monster game he had 2 sacks which is very out of the ordinary for him. I think today was more a case of tennessee playing poorly with backups and injured players than some kind of revelation of how great harris and branch are.
But I would gladly welcome being proven wrong, let's see if they can keep it up against bettter competition.
Unfortunately you are right about his college career and you'd think it would be a bad sign for his pro career. BTW he actually went up against a backup last week too and did nothing. He has a good first step but gets consistently pushed up field out of the play. The only thing left to hope for is that when he develops an inside move he can exploit tackles more as they have to honor the inside, making his outside step more effective.
That's a lot of number crunching, done for no real reason. PFF isn't interested in correlation of their tariff to win percentage. Their rating is to actually attempt to look at how a QB actually played, regardless of the rating. Cutler today, for instance, let's say a couple of those crappy passes were caught, could have boosted his rating. PFF would look at it, and give him a low grade, because, although the passes were caught, they were bad passes that receivers made good or great plays on to catch. Just the same, a QB who has a high rating, but throws a bunch of 5 yard passes might get a lower grade, because he didn't really do anything special. Another QB, with a low rating, might get a high PFF grade because he made great throws, but receivers didn't catch them.
Comparing their grades to win percent is pointless.
lol resnor that post was from almost 5 weeks ago. Are we regurgitating this discussion again?
I know you, phins18 and some others just can't understand this, but the whole point of any player evaluation should be to determine how well that player did on things that matter most for winning. You don't want to be like Philbin and focus on something like tidiness if one can't demonstrate that helps increase win%.
That's precisely why correlation (which you need for causation, unless there's a causal factor directly counteracting that) to win% is the key metric. And whatever potential advantages you think looking at tape has, if you can't demonstrate your method allows you to predict outcomes better, or correlate with win% better (because that's what matters!), then you simply don't have as good a method as a simpler, stupider method like passer rating.
In any case, don't forget that the utility of passer rating (and most stats) only show themselves over time. I've said over and over again that for a single play or for small sample size you don't rely on stats as much as looking at tape. But over time it's almost always better than human memory.
Regardless.. no point regurgitating this again. The basic logic of my argument isn't specific to sports but is general to the application of statistics in science, engineering, medicine, etc.. (would love to see you try and make this type of argument for rejecting clinical trials!) and nothing said here will change that. So if you don't understand it, no harm caused.
You're assuming that PFF takes all their grades and crunches the numbers to come up with who is going to win.
Do you know they do that or are there predictions, opinions?
Read post #243 again. There are two parts to that post. First part deals with predictions pff made (win totals), second part deals with post-hoc ratings for QB's. For predictions I compared to Vegas, for post-hoc ratings I compared to post-hoc stats, like passer rating.
But we already know there's a correlation to passing and winning. That doesn't tell us that PFF's assessment of a given play is properly attributed to the right player in a passing situation (or any situation for that matter).
Again, that is a leap in logic.
Like I said.. read post #243 again. I didn't criticize pff's evaluation of a single play.
I looked at pff's QB ratings and compared them to passer rating. The question was which approach captures more of what matters for winning. And absent evidence of a causal factor undermining the correlation to win%, correlation to win% is the way to go (which is why other football stats sites do this, and it's also why moneyball did it, etc..)
You know, you do this read post #XXX a lot.
I did read it. You started with:
Comparing their predictions to their ratings is a poor way to determine if their ratings are good (read posts #s 250, 252, 254) for the reasons resnor and I have already said.
I hadn't seen it until tonight.
Regardless, the comparison you did is severely flawed. PFF isn't interested in the actual outcome of the game, i.e., win/loss. They're interested in how a particular player did, irrespective of outcome. It's a simple concept that you, and others on here, ignore, because there is this belief that QB play is mostly responsible for wins, the simple concept being that almost everything a QB does is predicated upon like 6 other players doing their job. Five blockers and at least one receiver. PFF is trying to grade what the QB does on his own, something passer rating cannot, does not, and will never do.
Yeah guys, like I said you just can't understand this. Won't stop me from doing such analysis in the future.
Anyway, time to move on.
Again, you seem incapable of understanding that a good passer rating DOES NOT mean the QB actually played that well. It MIGHT mean that, but it MIGHT also mean that other players played great, and the QB benefited. Rating is relatively meaningless, in and of itself, to tell how the QB actually played.
lol. Sure thing, you say X, try to prove it incorrectly, then defend it by saying you tried to prove Y and then say we don't understand and you're not talking about it anymore.
Right. We disagree with your comparison, so we must not understand it.
It is clear that you don't understand why the two things you compared are in actuality completely different, and aren't comparable in the least.
The owner of PFF (before Colinsworth bought them) was on local gambling radio show about 2-3 years ago. It was when Rodgers had that 4 TD performance but PFF had him rated super low for that game and people were complaining. The owner explained that their grades are based on expectations of the play for each position. So when a QB is on his half of the field, the expectation is that he throws routes that move the chains to get the team in position to score (the RZ). The expectation for a QB to score doesn't happen until you approach the RZ. So the reason why Rodgers was graded low that night, was because I believe all 4 of those TDs were long with his WRs doing the majority of the work.
I'm not really looking to get into these statistical gymnastic arguments that so many here love, but I just wanted to pass on information that I heard from the PFF owner on how they grade.
The announcer Chris Collinsworth? Because I think he's one of the smartest guys in football. He actually spends about 60 hours each week studying both teams before his Sunday broadcasts- I have massive respect for him and what he does.
It doesn't have anything to do with the conversation here, just curious.
Yes, that is correct.
Harris playing better...good job... make us proud and apologize for wanting to take someone else..
Harris has had better production then taco and Derek Barnett.
He had played three games up until this point. THREE.
Some of you people are insane.
And to say he's done nothing up until the point is wrong. He played very well in London too. I know it may be hard to believe because the stat sheet said zero sacks, but I promise you he did.
By the way, today was a pretty good indication as to why we took him in the first round. He looks a lot like Cam Wake with the way he explodes off the line, uses his hands, and bends his body around the blocker.
He's going to be a damn good player.
He played three regular season games and 4 preseason games before this game getting significant reps in all of them and did nothing. Yes he did have a QB hit last game and a tackle for loss but he was playing against a backup, he was also playing against a backup and hurt player in this game. All of his college production was against scrub tackles as well whenever he played against decent competition he did nothing. When he plays well against good competition let me know and I will gladly eat humble pie.
He's actually played REALLY well over the last 2 games. He's been disruptive and never stops.
Of course he's a rookie and most intelligent/objective people aren't expecting him to play like Jason Taylor or Bruce Smith yet.
I've noticed him the last couple of games. People forget that JT wasn't JT either as a rookie. In fact, the vast majority of rookies take time to acclimate, even the ones that go on to HOF careers. I'm not predicting Harris will be inducted into the HOF or anything at this point, but it is beyond ignorance to roast a rookie for not dominating right out of the gate. The "he is a first round pick and should start right away" quote is on my list of the dumbest things that fans say every year.
I'm going to make your post my new sig...lol
There is a good bit of jibberish involved in this thread.
Ok, I hope so...
Yeah...Harris SUCKS!! And so does Timmons.
Who ever said timmons sucks? And harris has not been doing much to warrant a first round pick so far. There are third and fourth rounders from this year's draft that are outplaying him by a mile. Shut your pie hole.
Are you crying? Guilty of something? If not, then why take this post so personal?
Your lack of football acumen is why your posts suck. Harris, especially in the last 3 games, has looked every bit of a 1st round caliber NFL football player. The problem is, you weren't attracted to him from the get go and so you feel slighted. Your "feelings" (wink*wink*) for someone else have clouded your judgment. Well, that and your overall ignorance of football.
I will say, however, that for a jilted lover of another draftee, you're much less violent than most would be.
Now, go sit down, crybaby.
Timmons is gonna be a in the pro bowl...shot at all pro..fu## the steelers.
I have been very impressed with Harris so far. He has shown some bite and looks way more athletic than his combine results. Hopefully he hasn't hit his ceiling, however, at this point I am excited for his potential.
No way deej...Timmons is washed up or else the Steelers would have kept him. haha
Calling Harris a bust before he even played a down...hahha
How's that run "D" doing with Hayes, Timmons, and Harris? Three players you disparaged. haha
I never disparaged hayes you imbecil, get your facts straight, and I didnt say timmons sucked, I just said he is a stop gap measure who would give us one good season before we had to find another solution.
As for harris he has not been very impressive so far, he has played like what I would expect a second or third round pick to perform like. Like I said there are third and 4th round de's from this years draft that are outplaying him.