DELJZC QB Rating Analysis (2014 Season)

As some of you know, I created my own passer rating formula about four seasons ago.

If you want specifics see here: http://deljzc.blogspot.com/2011/08/correcting-qb-rating-system.html

Since I believe the most relevant "factor" in a season is Passer Rating For vs. Passer Rating Against (i.e. Passer Rating Differential), I will track and follow how we are doing.


Week 1 vs. CLEVELAND

Roethlisberger DQR: 127.0

Hoyer DQR: 105.1


SEASON TO DATE:

Offense DQR: 127.0

Defense DQR: 105.1


Difference +21.9

 

That looks solid and I like your formula. Thanks for the hard work.

 

Week 2 @ BALTIMORE

Roethlisberger DQR: 47.4

Flacco DQR: 141.0


SEASON TO DATE:

Offense DQR: 89.9

Defense DQR: 120.8


​Difference: -30.9

 

That is embarassing, especially considering Flacco really isn't that good.

 

Week 3 @ Carolina

Roethlisberger DQR: 133.9
(Key stats: 22/30 for 196 yards, 1 sack, 0 TO's, 9 drives/30 pts., 2 20+ yard plays)

Cam Newton DQR: 70.0
D. Anderson DQR: 200.0
(Key stats: 31/43 for 367 yards, 3 sacks, 1 TO, 9 drives/19 pts., 3 20+ yard plays)


SEASON TO DATE:

Offense DQR: 102.6

Defense DQR: 113.4


​Difference: -10.8

 

Ben likes to chuck the ball a lot. Most of the offensive production came from the run game. I wonder if Ben will yell to get more passes in.

Or... I'll ask a question: Do you think there's enough ball to go around for both a franchise QB with two SB rings and two RBs that can put up 100+ yards/game?

 

All comes down to how well the line plays. They keep everyone clean, and they're be plenty of balls for everyone

 

Keep up the good work Del.

 

Provided they don't go 3 and out all the time, and provided the defense can get the other guys off the field - yes, more than enough ball.

 

Hi Del,

Read through your explanation on your blog. I realize the R/R stat is not weighted as highly but I was left with the feeling that it's already accounted for by the others (esp in PPP since, in the end, the R/R behavior is only valuable if it either allows the good guys to score or, in the case of the differential, keeps the bad guys from scoring because they are pinned back).

Having said that... the R/R stat *does* represent a more exciting product to watch.

Anyway... you're more in touch with this stuff than me so I defer to your judgement (never mind your right to post )... just thought I'd ask about your philosophy since you've had this bake for a couple seasons now.

 

Del, that is a nicely done.. must have taken a bit of time. Good job.

I am currently thinking if we ever have a disagreement about ideas.. I lost.

 

I sure hope so. Nothing like killing opposing teams both on the ground and in the air.

 

Sacks are not counted in any of the other statistical categories except as "pass attempts". I purposely do NOT count negative yardage from sacks in "Total Yards per Pass Play". I've thought about it, but don't want sacks to effect things too much (especially since they weren't even considered at all in the old formula).

I agree sacks do effect points per possession. Lots of things do and that's the real big change of my formula vs. the traditional passer rating stat. I mean when Bell runs for 81 yards, or we get a turnover that puts us on the opposing 10 yard line, that helps Roethlisberger's DQR because those drives will likely result in points.

I guess I feel over the course of a season, those "starting field position" issues balance out (i.e. there are also time Roethlisberger comes on the field at his own 5-10 yard line to start a drive as well).

The purpose of the rick/reward category was to create a minor stat that rewards success by a quarterback playing ANY STYLE of passing game. There are coordinators and coaches that accept sacks with the understand the design of the play was to pass deep. There are also coordinators and coaches that accept dink-and-dunk passing from their quarterback in exchange to avoid sacks, hits, etc. The risk/reward category hopefully rewards both styles equally and fairly.

The risk/reward category also allows maybe some idea to why some quarterbacks in the 1960's and 1970's were much more successful than what the traditional passer would indicate.

 

Since that article I have decided to only take out quarterback sneaks and kneel downs. There aren't as many "designed" option runs as you think and I have been rewarding them to quarterbacks that do them well.

If I look at year-end stats on quarterbacks, I normally just remove 10-16 rushing attempts to compensate for kneel downs and sneaks and remove ZERO yards (since kneel downs normally have -1 yard rushing and sneaks have +1 yard rushing). All the other yards and attempts I count.

 

You know Ben should be thankful for the running game. It was a pure showing of our will. If that continues, Ben's passing stats will be under 20 throws a game, but with 200 yards and 2 TDs every week.

 

I think my formula could be more easily explained as follows:

Quarterbacks can gain DQR rating points from the following five statistics:

1. Completion Percentage (Pass Completions/Pass Attempts)
2. Total Positive Yards per Pass Play [(Pass Yard + Rush Yards)/(Pass Att + Rush Att + Sacks)]
(as stated above QB sneaks and kneel downs are excluded)
3. Points Scored per Possession by a QB's offense
4. Turnovers (Int + Fumbles Lost) per Pass Play
5. Risk/Reward Factor = [(20+ yard plays - Sacks)/Pass Plays]

Now let's look at it in actual points available to score:

Completion Percentage

Worst QB (55.8%) = 13.9 DQR Rating Points
Average QB (61.2%) = 17.5 DQR Rating Points
Best QB (69.5%) = 23.0 DQR Rating Points
Perfect QB (76.3%+) = 27.5 DQR Rating Points

Yards per Pass Play

Worst QB (5.63) = 13.6 DQR Rating Points
Average QB (6.73) = 22.7 DQR Rating Points
Best QB (8.09) = 34.1 DQR Rating Points
Perfect QB (12.0+) = 66.7 DQR Rating Points

Points per Possession

Worst QB (1.32) = 5.3 DQR Rating Points
Average QB (1.86) = 14.3 DQR Rating Points
Best QB (2.96) = 32.7 DQR Rating Points
Perfect QB (3.50+) = 41.7 DQR Rating Points

Turnovers/Pass Play

Worst QB (4.9%) = 8.2 DQR Rating Points
Average QB (2.8%) = 20.4 DQR Rating Points
Best QB (1.0%) = 30.7 DQR Rating Points
Perfect QB (0.0%) = 36.7 DQR Rating Points

Risk/Reward Factor

Worst QB (-2.6%) = 1.9 DQR Rating Points
Average QB (+2.0%) = 10.9 DQR Rating Points
Best QB (+7.4%) = 21.8 DQR Rating Points
Perfect QB (+10%) = 27.5 DQR Points

TOTALS

Worst QB = 42.9 DQR Rating*
Average QB = 85.8 DQR Rating
Best QB = 142.3 DQR Rating*
Perfect QB = 200.0 DQR Rating

* Note this is not an actual quarterback last year but only if the same quarterback led the league is all the best and worst categories (that is not the case). Last season the best QB was Peyton Manning (133.6 DQR) and the worst "starter" was Geno Smith (56.6 DQR)

 

Week 4 vs. Tampa Bay

Roethlisberger's DQR: 98.9

M. Glennon's DQR: 114.6


Season to Date:

Roethlisberger DQR: 101.2

Defense DQR: 112.9



​Season Differential: -11.7

 

Week 5 @ Jacksonville

Roethlisberger's DQR: 81.6

B. Bortles' DQR: 39.2


Season to Date:

Roethlisberger DQR: 97.9

Opponent's DQR: 97.7

Season Differential: +0.2

 

This puts Ben at slightly above average, which is what he is. So much for the "Elite" discussions? This also leaves me with my belief that we let him run out the last 2 years of his contract then tag him if he is still worth it. I suspect he would pout and throw a hissy fit if the Steelers did that.

 

I think Ben is clutch ,but not elite.... I think he could become elite if he keeps working at his craft....IMO Ben has never been a true student of the game , but he is getting better running the no huddle so I think he has been working on ore film study ....

 

Err, I think this means he just eeked out against our competition... which sounds about right. Way too few datapoints to have conversations about below/above average... we don't have enough other QBs/teams in the calculations.

I dunno how easy it is for DelJZC to run his analysis for all matchups and see the season differential for all QB's that might be interesting to see how the teams stack up. (having said that, given DelJZC's recently voiced frustrations with the team and coaching, I'm not sure how much effort he wants to put in )

 

I am debating changing the formula involved with points per possession.

I am worried it currently has too much effect on quarterback differences (not that it's not important) but I don't want the entire DQR rating to so closing match one statistic (and it's starting to).

I have Roethlisberger's "estimated" DQR every year of his career and I have a much more accurate (using game data) of his DQR every year since 2010.

I have estimates (based on season stats and some assumptions on sneaks/kneel downs) of every major starting quarterback since 2011.

 

For the math guru's out there, the problem with the points per possession statistic is the "slope" of the formula is too steep (thus too great a DQR differential for different PPP scores).

Each of the formulas above is a linear equation (i.e. straight lines).

The flatter the slope, the less point differential between the best and worst in a specific category.

The easy way to fix this is to change the formula for PPP from:

(PPP-1.0)*100/6

to

[(PPP-.75)/1.5]*100/6

I can also easily keep the "perfect rating" at 200 if I just change what constitutes a "perfect game" from 3.5 PPP to 4.5 PPP. Most of the time this happens anyhow.

It's was hard already to make a perfect game and I'm making it harder, but I think in total evaluation what you'll see if "perfect 1st halves" then the teams take the pedal off the floor since they are already out to big leads.

Plus, if you like the idea that perfect games should be very, very rare then this won't bother you.