A lot of the reason QBR was created is cause of how imperfect the old passer rating was.
While passer rating is hardly perfect it is a hell of a lot better than QBR and one of the best stats overall to measure a QB by in the long term when compared within era.
If you were to go by something other than Passer Rating that is in the same ball park I'd pick "Real QB rating" on CHFF cause it is easy to understand and while imperfect as well is reasonably fair.
Is there somewhere where the formula is published? It looks good to the eye test and I agree with the inclusions of the stats added to the traditional QB rating, but I'd like to see the formula.
There's always the urge to improve the traditional QB rating as it leaves out important things. Incompletions hurt your rating but sacks don't affect it at all, for instance. The CHFF Real QB rating seems to address some of that, as they say "Real Quarterback Rating includes rushing attempts, rushing yards, rushing TDs, fumbles and sacks to produce a new kind of rating that measures a quarterback’s overall performance with the ball, not just as a passer when he actually releases the ball (which is all that passer rating currently measures)."
But as I can't find the rating, I can't tell if their "Advances" category which simply lists "completions + rush attempts" includes completions and rush attempts for no gain. Should that be an "advance"?
Here's my proposal on how we might tweak the passer rating.
1. Completion % becomes Positive Play %
The percentage of QB dropbacks and rushing attempts that results in the gain of at least one yard. Mesaures the basic competence of a QB to execute the bare minimum of offense.
2. Yards/Attempt to include Rushing attempts and sacks.
Pretty much same as the CHFF rating, as far as I can tell. Measures the ability of the QB to gain large amounts of yardage consistently. It of course credits the QB for the receiver's YAC, but unless you can track what % of receivers' yardage comes after first contact, we aren't going to be able to accurately assess it. If Brady throws a screen to Lewis at the LOS and he goes 5 yards untouched before being tackled at 10 yards, you might convince me that Brady only "deserves" 5 yards but you won't convince me that Brady doesn't deserve any yards. Watching Brady after Bledsoe we know that accurate short passing is an actual difference-making skill that you ignore at your peril.
3. TD % becomes TD + First Down %
TDs are wonderful, the goal of any offense, but gaining first downs is a skill that's ignored by traditional passer rating. The previous category rewards gaining yards in large chunks. This category rewards consistently moving the chains and then cashing in at the end. If you want to weight TD passes more than getting a first down, I won't argue, but if a QB converts six first downs in the air before handing off to a back for a 1 yard score, shouldn't the QB get some bonus for that?
4. INT% becomes INT + Fumble %
Also pretty much same as the CHFF rating, as far as I can tell. Again I need to see the formula. If there's a way to include dropped interceptions and Fumbles recovered by the offense, I'd do that as well as there is no QB skill to having a defense drop a sure INT or fail to recover a fumble.
PLUS: Garbage time cutoff. The recent discussion on garbage time scoring and the prevent defense makes me think that the same idea should be applied to QB rating. If we could disallow any stat accrued when the win probability is above 99% for one team, that would eliminate garbage time stat padding.
I think that would improve even the CHFF rating. Still, traditional passer rating is usually fine most of the time. Certainly better than any scheme that tries to apply subjective weights and judgement to stats like QBR and CHFF. Football Outsider's DVOA ratings, as far as I understand it, are entirely stat based and pass the eye test much better than QBR though I couldn't explain it fully. I think it just asks how a QB did vs. every other QB in the same situation, no matter how it happened.
