QBR

[Kelly the Dog]

Found this article on the old passer rating.

 

 

 

Passer Rating Differential is not Passer Rating. All it means is you better have a good pass defense as well as throw the ball. Saying that 90% of winners are good at passing and also good at stopping the pass is not an alarming stat in any manner. It's exactly to be expected.

[San Jose Bills Fan]

I'd like to point something out. Those defending the old passer rating system seem to be relying on its efficacy in predicting winners. Fine, it predicts winners. That isn't the point of the statistic. The point is to specifically identify the role of the quarterback in his team's success relative to his running backs, receivers, tight ends, etc.

 

Team batting average is great at predicting winners in baseball too, but sabermetricians dismissed it long ago in the context of identifying individual achievement at the plate, because it isn't comprehensive. It's the same with passer rating. It isn't a fallacious statistic, but it is rather arbitrary. Taking 4 stats, weighing them, then smushing them together isn't comprehensive enough in an age when so much other information is available.

 

Take the following example. In passer rating, if the QB dumps the ball off to his running back, who jukes two defenders while using aggressive second-level blocking by a FB and OT and scores a 50 yard TD, the QB gets the exact same amount of credit as he would if he placed a perfect bomb to a streaking WR in single coverage for a TD. In my opinion, you simply can't defend a statistic that is so poor at accounting for context.

Well Sage, the counter-argument would be that in the long run, over a large sample size, that those things would even out/cancel out and that the introduction of additional data doesn't necessarily make for a more accurate barometer… in other words, more is not necessarily better.

 

I don't necessarily believe that to be the case but I imagine that that would be the counter-argument.

 

Anyways it's great to see that Ken is actually involved in Sabremetrics and I'm looking forward to hearing what he has to say about all of this.

 

 

[KRC]

I agree with some of the comments here regarding SABR-metrics and its application to football. I have always said that stats without context are meaningless. You can try to apply stats to the discussion and you can continue to analyze players using revised stats, but you will always be left with intangibles associated with a team game.

 

Next on my list is an overall offensive rating and defensive rating. Now that we have the rusher rating and passer rating, we can use those concepts to create the overall offensive and defensive ratings. I think that is a better indicator than the individual rusher and passer ratings, as it takes into account team play and not individual play.

 

It is going to take time and you will never find the perfect formula. However, that should never stop you from trying to get a better handle on what is happening on the field. Just never lose sight of the fact that stats will never tell the whole story. It doesn't happen in baseball and it will never happen in football.

 

Wow Ken. Great work and I didn't know you were in the Cold, Hard, Football Facts group. Seems like it would be interesting work and fun, when the results are arrived at.

 

I really like your Rusher Rating System and the idea of normalizing stats across years to reduce the apples vs oranges aspect. I think it does a good job of trying to tackle the problem of statistical relativism across time.

 

I was surprised by how often Sayers, Simpson, and Dickerson fumbled the ball and agree wholeheartedly that this is an important measure of players at their position. That was the fatal flaw for each of those players but based on that small amount of data, it also seems like there's been a greater emphasis on "ball security" in today's NFL. Even Jim Brown, who was a man amongst boys at times had a fairly high fumble percentage although maybe not relative to his contemporaries. I wonder if your formula adequately accounts for this difference in eras and my other quibbling criticism is that I wish (because I'm lazy) you would have added columns for TD% and Fumble% just for easy reference.

 

The challenge for people like yourself I would imagine is how many different criteria to use in the formula. In a perfect world everything could be accounted for but I think also at some point that becomes totally impractical and impossible.

 

And even if it is possible, it yields very diminshed returns. I'm talking about things like how good that RB's (or QB's) offensive line was or how good their defense was because all those things have some bearing.

 

Anyways, the system that you personally devised is a great way of looking at running backs and I'm looking forward to the more comprehensive rating which includes pass receiving stats for the running backs.

 

It will enable us to compare guys like Thurman Thomas, Roger Craig, and Marcus Allen… who didn't make the cut in your initial rating. Also, I'm thinking that when you take receptions into account that Marshall Faulk will be at or near the top.

 

Great stuff, Ken, and thanks for sharing.

 

The items included in the tables were an editorial decision by Kerry Byrne (the guy who runs CHFF). I gave him all of the raw data.

 

This is just the first stab at creating a rusher rating system. It will continue to evolve over time and other aspects (like the receiving aspect) will be added and tweaked. As I mentioned in my other post, it can also start to give us an overall offensive and defensive rating, which I think will be a better indicator since it involves the "team" and not the "individual."

 

Thank you for the comments.

[Buffalo Barbarian]

The only statistic that matters is the W

Beat me to it.

[SageAgainstTheMachine]

Beat me to it.

 

Advanced statistics can aid general managers in building teams that win. Ergo, they matter.

[San Jose Bills Fan]

The only statistic that matters is the W

I understand what you're saying but nothing is quite that simple, either.

 

The compilation and use of data will only increase as we "progress" as a civilization.