Total QBR (testing...)

Discussion in 'Second to None' started by chance17, Nov 29, 2012.

  1. chance17

    chance17 Walk On

    Joined:
    May 8, 2012
    Messages:
    1,507
    [​IMG]
    I decided to try my hand at making a total quarterback-rating (QBR), but not as secretive and confusing as ESPN's proprietary formula.

    I preface this with two things to keep in mind:
    1) It's a game and it's supposed to be all in good fun
    2) The saying goes... There are three kinds of lies: lies, damn lies, and statistics.

    Let's start off with a primer on passing efficiency and what total QBR is meant to do.



    Passing efficiency

    A tried and true formula that gives a number to slap on a QB's performance. Depending on if you follow the NFL or NCAA, it has a couple of different forms, but both factor in the following: passing attempts, completions, passing yards, touchdowns, and interceptions. Why? Because they are the most important and easy stats to deal with and Elias sports bureau kept these stats when this was created.

    For all the number whores out there...

    a=((Comp/Att) -0.3) / 0.2
    b=((Yards/Att) - 3) / 4
    c=(TDs/Att) / .05
    d=(.095 - (Int/Att)) / .04

    a, b, c, and d have minimum values of 0 and maximum values of 2.375

    Quarterback Rating = (A + B + C + D) / .06
    [ { (8.4 * yards) + (330 * touchdowns) - (200 * interceptions) + (100 * completions) } / attempts ]

    The main gripes that people have about these are the arbitrary scales (Between 0 & 158.3 for NFL and -731.6 & 1261.6 for NCAA :confused:), it doesn't take into account rushing statistics, and it doesn't factor in winning and intangible qualities (clutch aka comeback kid). Arbitrary scales lead to confusion, especially when a 100 QBR in the NFL is great and in NCAA not so much. Rushing is an important part of the game, especially at the collegiate level where defenses aren't as talented. Winning is everything, or at least should factor into our equation somewhat as to not reward garbage time as much as play that results in a win. Also, the NFL formula is based on very old (1971) statistical averages and probably should be updated to compensate for increased passing stats.




    ESPN's Total QBR

    This was recently developed and is being used by ESPN with a lot of backlash from fans and Aaron Rogers. Its a great concept that basically can be summed up as how much does a QB give his team a chance to score when he makes a play with the ball. Using stats that nobody else has (e.g. yards a pass is in the air, percentages relative to down and distance, as well as quarter, and score) the numbers get sent into a magical equation that, if ever written out, would instantly blow the minds of all who beheld it. It has the benefit of being on a 100 pt absolute scale. Other than that, it takes into account all the things everyone wants to believe matters, but nobody can really put stock in it because they don't understand it.





    My Total QBR

    Update the NFL formula for passing and add to that rushing attempts, yards, touchdowns, fumbles, and wins. Keeping the NFL-style formula, while more complicated, allows for a minimum and maximum on each value. This helps out pocket passers and scramblers both because their potentially negative values from too many interceptions or sacks taken (rushing attempts in NCAA) can be negated. Also, video games allow for big plays to be the norm sometimes, especially when running with the QB, so it prevents bogus inflation from option QBs with big averages.

    In general, these things give a QB points...
    - Passing for 3.3 yds per attempt (consistently getting first downs)
    - Higher than 50% completion rate
    - More TDs per attempt
    - Fewer INTs per attempt (less than 9.5% of attempts)
    - Running for more than 3.5 yds per carry
    - More TDs per carry
    - Fewer Fumbles (less than 5% of carries)
    - Win percentage (above 50% is positive)

    a=((Comp/Att) -0.5) / 0.1
    b=((Yards/Att) - 10/3) / 4
    c=(TDs/Att) * 20
    d=(.095 - (Int/Att)) * 25
    e=((RYds/Carry) - 3.5) / 4
    f=(RTD/Carry) * 20
    g=(.05 - (Fum/Carry)) * 40
    h=((Wins/Games) - 0.5) * 4

    a,b,....., and h have an imposed minimum of 0 and maximum of 2.375

    Total QBR = (A + B + C + D + E + F + G + H) * 6.25


    These numbers should keep us roughly on a scale like total QBR where scores up to 40 are low performing QBs, above 60 is good, and above 75 would be stellar in real life, but attainable on a game. Here are a few college careers and seasons I picked out to demonstrate with (fumbles weren't listed, so these guys get a perfect score on that).

    Player Total QBR[/FONT]
    Timmy Chang 45.69[/FONT][/FONT]
    Tony Rice 47.52[/FONT]
    Eli Manning 52.62[/FONT]
    Eric Crouch '01 57.30[/FONT]
    Chris Weinke '00 58.48[/FONT]
    Geno Smith '12* 60.42[/FONT][/I]
    Doug Flutie '84 61.27[/FONT]
    Jim McMahon 61.47[/FONT][/FONT]
    Peyton Manning '97 63.14[/FONT][/FONT]
    Steve McNair '94 66.88[/FONT][/FONT]
    Kordell Stewart '94 66.90[/FONT]
    Michael Vick 66.94[/FONT]
    Andrew Luck 69.46[/FONT]
    Johnny Manziel '12* 70.82[/FONT][/I]
    AJ McCarron '12* 71.65[/FONT][/I][/FONT]
    Kellen Moore 71.18[/FONT][/FONT]
    Charlie Ward '93 74.96[/FONT]
    Aaron Murray '12* 74.98[/FONT][/I]
    Tim Tebow 76.33[/FONT][/FONT]
    Colt Brennan '06 76.99[/FONT][/FONT]
    Graham Harrell 77.21[/FONT]
    [/FONT]
    * - denotes season in progress


    Things to take away from that:

    - Hard to break 60 IRL
    - Above 70 is insane
    - Doesn't seem to be skewed for or against scramblers or pocket passers particularly
    - Modern QBs are "better" (college offenses are vastly improved and defenses cannot keep up because of modern schemes and rule changes)




    Drum Roll Please...

    How does your QB stack up in terms of total QBR?

    Here are the results through week 7:


    Player Team Coach Total QBR[/FONT]
    Laurence Hawkins [​IMG] webbs6212 22.72[/FONT][/FONT]
    Blake King [​IMG][/FONT] T3AM_F1GHT_0N [/FONT][/FONT]31.29[/FONT]
    Ryan Peterson [​IMG][/FONT] redwings8831 [/FONT][/FONT]32.61[/FONT]
    Corey Arnold [​IMG][/FONT] DrunkenSoldier [/FONT][/FONT]35.09[/FONT]
    Brandon Davis [​IMG][/FONT] Shnoogans10 [/FONT][/FONT]39.80[/FONT]
    Gunner Kiel [​IMG] nautica[/FONT][/FONT] 54.02[/FONT][/COLOR]
    Daniel Bowen [​IMG] [/FONT][/FONT]broth262 58.94[/FONT][/COLOR]
    Stephen Green [​IMG][/FONT][/COLOR] MarkyMark67 [/FONT][/FONT]69.16[/FONT][/COLOR][/FONT]
    Jameis Winston [​IMG][/FONT][/COLOR] KnightNoles [/FONT][/FONT]71.723[/FONT][/COLOR][/FONT]
    Matthew Patterson [​IMG][/FONT][/COLOR] Coach_III [/FONT][/FONT]78.86[/FONT][/COLOR][/FONT]
    Raymond Keith [​IMG][/FONT][/COLOR] [/FONT][/FONT][/B]trevorg234 [/FONT][/FONT][/B]84.10[/FONT][/COLOR]
    Daniel Bryant [​IMG][/FONT][/COLOR] chance17 [/FONT][/FONT]85.05



    The season's not over yet, so some of these guys may see their numbers balance out toward the middle. Two of the Heisman-hopefuls get love from this rating (Patterson and Bryant), but Bowen gets snuffed because of lower completion%, yards per carry, and winning percentage. For video game stats, i think this works okay. I may tweak it some more and I'm open to suggestions. What do you guys think? I'm quite fond of it given that my QB is killing it this year. :cool:
     
    • Like Like x 7
  2. KnightNoles

    KnightNoles Learn to Compete

    Joined:
    Jul 6, 2009
    Messages:
    19,821
    wow bravo!
     
  3. KnightNoles

    KnightNoles Learn to Compete

    Joined:
    Jul 6, 2009
    Messages:
    19,821
     
  4. Shaun Mason

    Shaun Mason Somebody you used to know.

    Joined:
    Feb 9, 2009
    Messages:
    24,887
    Featured Threads:
    5
    Curious how that stacks up for my 10 minutes of trying, LOL.
     
  5. chance17

    chance17 Walk On

    Joined:
    May 8, 2012
    Messages:
    1,507
    Just tried what you have listed, but the numbers range from 0.42-10.26.

    I like having completion percentage and winning percentage in there ;)
     
  6. Shaun Mason

    Shaun Mason Somebody you used to know.

    Joined:
    Feb 9, 2009
    Messages:
    24,887
    Featured Threads:
    5
    Completion percentage is irrelevant to your effectiveness, and you can perform well individually and still lose.
     
  7. chance17

    chance17 Walk On

    Joined:
    May 8, 2012
    Messages:
    1,507
    [​IMG]

    The point was to make something like the ESPN total QBR, but have it be able to be calculated from regular stats. This was my first attempt at this and most of these numbers are arbitrary and will change as we see individual performances get assessed. I mostly like the idea of being close to a 100 point scale and it would be hard as hell to get a 100 with this. :D
     
    • Like Like x 1
  8. chance17

    chance17 Walk On

    Joined:
    May 8, 2012
    Messages:
    1,507
    Just a snippet here for S2N...

    Week 7 Total QBR

    Player Team Total QBR[/FONT]
    Ryan Peterson Wisc 30.55[/FONT]
    Corey Arnold LSU 32.30[/FONT]
    Christian Suntrop BC 39.06[/FONT]
    Blake King VT 41.28[/FONT]
    Jameis Winston [/COLOR]FSU 68.81[/COLOR][/FONT]
    Stephen Green Miami[/COLOR] 69.53[/COLOR][/FONT]
    Raymond Keith Illini[/COLOR] [/B]87.20[/COLOR][/FONT]
    Matthew Patterson [/COLOR]UGA 91.31[/COLOR][/FONT]
    Daniel Bryant SCar[/COLOR] 92.61[/FONT]
     
    • Like Like x 2
  9. chance17

    chance17 Walk On

    Joined:
    May 8, 2012
    Messages:
    1,507
    This formula hasn't produced any glaring flaws, until now. Looking at week 8's stats and the performances I don't think the numbers are reflecting player performance that well.

    I'm going to take the formula and tweak it to weight the passing and rushing relative to the number of "action plays" from each category. I am also considering subtracting sacks from the rushing attempt numbers so that taking more sacks doesn't make you look like a "runner" according to this formula.

    I will update the previous posts after some number crunching late tonight. Have to go stain the deck (I know, South Carolina weather...).
     
    • Like Like x 2

Share This Page