QB Passer Rating

[Don Otreply]

It is arbitrary and capricious, and makes little sense,  a perfect game by a QB would need to be 100 percent completion rate, no fumbles or negative plays by said QB, and a victory in said game. This is what a perfect QB rating actually is. 

[Punch]

1 hour ago, GoBills808 said:

to illustrate why passer rating is hopelessly flawed and too heavily weights completions

 

QB1: 20/40 for 600 yards no TDs or INTS- passer rating of 95.8

QB2: 20/20 for 150 yards no TDs or INTs- passer rating of 97.9

 

passer rating basically says the inherent value of a completion (for zero yards mind you) is still somehow worth 22.5 yards lol...QB1 is getting 15 yards every time he throws the ball while QB2 gets 7.5

More than a decade on and we're still debating JP Losman vs. Trent Edwards.

[Matt_In_NH]

Passer rating is a flawed stat for many reasons.  It does not consider many important aspects of playing QB and gives full credit once you achieve a certain level.   I view it as one thing to look at just like looking at TD/INT ratio.  The intent of passer rating was to take all passing stats into account and while it is interesting, it just fails to do that all that well.   The ESPN QBR thing is also flawed.  The NFL should adopt something new.

[Shaw66]

Well, I don't understand it, and it's based on someone's subjective judgment of how important some stats are relevant to others, but in the end it works.  

 

It doesn't really matter that the maximum score is arbitrary and capped, because no one is putting up the maximum score for a season.  (Passer rating isn't all that useful as a one-game stat.)

 

What matters is that at the end of the season, the passer ratings actually do have the best throwers on top and the worst on the bottom, and it does that.   That is, the ratings correlate with our understanding of of who's best and who isn't.  

[Chaos]

15 minutes ago, Shaw66 said:

Passer rating isn't all that useful as a one-game stat.)

In individual games, there is a high level of correlation between winning and the team with the higher passer rating.  Win the passer rating battle, win the game.  This reflects both the QB and the defensive efforts. 
 

I am not sure any statistic comparing across single games is particularly useful, other than wins/losses

 

[Royale with Cheese]

Passer rating doesn't take in context.  That's the biggest flaw.

 

[Franco_92]

2 hours ago, GoBills808 said:

to illustrate why passer rating is hopelessly flawed and too heavily weights completions

 

QB1: 20/40 for 600 yards no TDs or INTS- passer rating of 95.8

QB2: 20/20 for 150 yards no TDs or INTs- passer rating of 97.9

 

passer rating basically says the inherent value of a completion (for zero yards mind you) is still somehow worth 22.5 yards lol...QB1 is getting 15 yards every time he throws the ball while QB2 gets 7.5

This kind of makes sense actually. If you are only hitting at a 50% rate and have SIX HUNDRED yards, your wide receivers are doing OBCSCENE work and you are being way too risky with that ball lol. Bringing in a hyper-unrealistic scenario doesn't really prove anything. The important thing to ask about a metric is, if you watch a bunch of games and understand its general score distribution, does it match up with what you saw from a quarterback in that game? Usually it does for me, at least it does it better than QBR which I believe is not even open-source. And it's not like anyone who hates passer rating has a better metric they point to and rely on. The reality is, any time there exists something that can theoretically be quantified in one easy number there will be humans who try to do it because it's a useful tool.

7 hours ago, GoBills808 said:

It's designed (back in the early 1970s mind you) so that each component would be constrained to 2.375 because they had set 1 as the baseline average QB performance...that measures out to a 66.7. That was an average QB performance in 1971 according to passer rating

Thanks- there was obviously an explanation for the "random" numbers, because nobody who develops metrics for anything just makes stuff up like OP implies. Cool to have the explanation, and that's usually the gist of it, shaping the distribution of performances so it is a bell curve around some mean to reflect the reality of the performances 

2 hours ago, Doc Brown said:

Some nerd should take all the different QB rating metrics out there (passer rating, pff grade, qbr, DVOA, EPA, etc.) and average it out to give the most accurate ranking.  To me the eye test of a skilled observer will always win out though.

Ah, but you've identified the snag with this method yourself 

[Dan Darragh]

It IS odd that if he had completed all 4 of his incomplete passes for touchdowns, it wouldn't have improved his score.

[Franco_92]

3 minutes ago, Dan Darragh said:

It IS odd that if he had completed all 4 of his incomplete passes for touchdowns, it wouldn't have improved his score.

Yea, but a decent metric is one for which breaking it requires incredibly unrealistic scenarios like the one you describe 

[The Wiz]

8 hours ago, sven233 said:

It really is a strange calculation.  During the "Perfect Game" against NE* in the Playoffs a couple years ago, he actually had a perfect rating until he threw his last TD pass.  Yup.....a completed TD pass actually brought his rating down.  I remember the discussions about this happening and yeah.....it's a very strange metric and calculation.

First thing that came to my mind when I started reading the OP.

 

He ended with 157.6 because of that pass.

[dave mcbride]

8 hours ago, GoBills808 said:

it has a ton of flaws

 

first of all it counts completions twice, so essentially passer rating says a completion is worth about 20 yards of offense when we know that most completions go for far less

 

it also penalizes interceptions too heavily...INTs are worth about -100yards according to passer rating and that's simply an inaccurate assessment considering a TD is only worth about 80

 

you like it and thats fine but saying it doesn't have any flaws is not correct

Team passer rating differential is the most predictive stat (outside of actual points scored and given up!) in all of football: https://www.si.com/more-sports/2011/06/23/most-importantstatpasserratingdifferential. Right now, the Bills have a team passer rating of 106.7 and opponents' collective team passer rating is 66.4. 

 

It's a good stat despite not being perfect. 

[Irv]

This should help:  

 

 

[dave mcbride]

1 hour ago, Chaos said:

In individual games, there is a high level of correlation between winning and the team with the higher passer rating.  Win the passer rating battle, win the game.  This reflects both the QB and the defensive efforts. 
 

I am not sure any statistic comparing across single games is particularly useful, other than wins/losses

 

Yup. Passer rating differential is the king of all NFL stats.

[JESSEFEFFER]

4 hours ago, Doc Brown said:

Some nerd should take all the different QB rating metrics out there (passer rating, pff grade, qbr, DVOA, EPA, etc.) and average it out to give the most accurate ranking.  To me the eye test of a skilled observer will always win out though.

The nerd's alias is Bruce Nolan and he named it QB Stew.

[FireChans]

8 hours ago, GoBills808 said:

a TD is always worth 6 points

 

an INT is NOT always worth -6 points, in fact EPA of an interception is nearer -4.5

So was the Jameis 30 TD 30 INT season actually not that bad lol

[Chaos]

4 hours ago, Doc Brown said:

Some nerd should take all the different QB rating metrics out there (passer rating, pff grade, qbr, DVOA, EPA, etc.) and average it out to give the most accurate ranking.  To me the eye test of a skilled observer will always win out though.

What are you evaluating with the “eye test”? A subjective judgement of who is the “best qb”? An objective judgement of who is the best passer?  or something else?

[Low Positive]

7 minutes ago, FireChans said:

So was the Jameis 30 TD 30 INT season actually not that bad lol

Actually, Jamesis had 33 TDs that season. That stat line is the most insane for a QB in my lifetime. 5,109 yards, 33 TDs and 30 INTs. Players have had more TDs of course, and even more INTs (Vinny Testaverde in 1988), but to have that much success and failure in a single season is really strange.

[dma0034]

The two major issues I have with Passer Rating are:

 

1. Doesn't take into account rushing TDs. Allen is punished for a 1 tard TD run but Mahomes can do a drop pass to Kelce who scores. Allen did more work but Mahomes gets a higher passer rating

 

2. Doesn't take into account situations. So if a QB is 10/20 with 0 TDs and 1 Int and losing 31-3 in the 4th he can improve quickly to something like 24/35 with 2 TDS and 1 Int when teams are plaing sidt coverage. Known in the Fantasy World as the Cutler Special

[MJS]

22 minutes ago, dma0034 said:

The two major issues I have with Passer Rating are:

 

1. Doesn't take into account rushing TDs. Allen is punished for a 1 tard TD run but Mahomes can do a drop pass to Kelce who scores. Allen did more work but Mahomes gets a higher passer rating

 

2. Doesn't take into account situations. So if a QB is 10/20 with 0 TDs and 1 Int and losing 31-3 in the 4th he can improve quickly to something like 24/35 with 2 TDS and 1 Int when teams are plaing sidt coverage. Known in the Fantasy World as the Cutler Special

Yeah, but it isn't called "QB performance rating", it is called "passer rating". It is only trying to measure passing success. It is one of many metrics to consider when evaluating the performance of a QB.

 

And you know, I'd say if you have a QB who is excellent at running, and horrible at passing, so that QB puts up rushing stats but has a low passer rating, that QB will eventually be exposed as a poor QB. You can't keep that up for a long time. Eventually, as a QB, you have to become a good passer, regardless of what you do on the ground.

 

QBR takes into account rushing stats, but that metric really has some issues, in my opinion.

 

In the end, the QB with the highest passer ratings through many games is almost always the better performing QB. It is a useful stat.

37 minutes ago, FireChans said:

So was the Jameis 30 TD 30 INT season actually not that bad lol

His passer rating was 84.3 that season. And you know? That seems about right.

 

Maybe there needs to also be some kind of volatility metric too. Taken with passer rating, you could see who is consistently performing without all the crazy highs and lows.

Edited October 3, 2023 by MJS

[FireChans]

4 minutes ago, MJS said:

Yeah, but it isn't called "QB performance rating", it is called "passer rating". It is only trying to measure passing success. It is one of many metrics to consider when evaluating the performance of a QB.

 

And you know, I'd say if you have a QB who is excellent at running, and horrible at passing, so that QB puts up rushing stats but has a low passer rating, that QB will eventually be exposed as a poor QB. You can't keep that up for a long time. Eventually, as a QB, you have to become a good passer, regardless of what you do on the ground.

 

QBR takes into account rushing stats, but that metric really has some issues, in my opinion.

 

In the end, the QB with the highest passer ratings through many games is almost always the better performing QB. It is a useful stat.

His passer rating was 84.3 that season. And you know? That seems about right.

 

Maybe there needs to also be some kind of volatility metric too. Taken with passer rating, you could see who is consistently performing without all the crazy highs and lows.

Passer rating, to me, has always served as a general “efficiency” metric. There’s an argument that completions are weighted too heavily, but as we know, most WR average over 10 yards per catch and running backs anywhere from 6 to 8, so completions usually equal extending drives. 
 

As with any efficiency metric in sports, there’s a way to game the system to inflate the metric (namely, checking down and taking safe completions). That doesn’t make the metric invalid, it makes it imperfect.