Passer rating

kevkev60614 · #1 10-01-2007, 02:15 PM

The current formula for passer rating in the NFL is:
=(MAX(MIN((COMP/ATT*100-30)/20,2.375),0)+MAX(MIN((YDS/ATT-3)/4,2.375),0)+MIN(TD/ATT*20,2.375)+MAX(2.375-INT/ATT*25,0))*100/6

Criticisms include:[*]It's too complex for the casual spectator to calculate.[*]The scale is arbitrary (out of 158.3).[*]The average is arbitrary (wikipedia says 78.9 for '00-'03).[*]Sacks/rushing yards aren't included.[*]The limits are arbitrary (77.5% completion percentage, 12.5 yards/attempt, 11.9% TD percentage, 9.5% INT percentage).[*]It sort of double counts completion percentage.

DVOA accounts for some of this but is a fairly complex calculation.

Why not use a formula such as this one I just made up:
=MIN(MAX((YDS+TD*30-INT*45)*8/ATT,0),100)

Based on 2006 data only:[*]It's a great deal simpler.[*]It's based on a range from 0-100.[*]The average for 2006 is ~50 (this will change over time, but whatever)[*]It has a (marginally) higher correlation with with winning.[*]The difference between team passer ratings has a (marginally) higher correlation with winning.[*]It's limited only by 0 and 100 overall (19 on each side in 2006 vs. 62 capped using current methodology).

Sacks/rushing yards still aren't included. But then again sacks are more often a function of line play than being a good passer, and one could argue that rushing yards have no business in a passer rating calculation anyway.

Again, DVOA is better than any number I could hope to come up with. But I think the NFL could use an easily calculated stat on a scale that makes sense for passer rating. Thoughts?

Jack of Arcades · #2 10-01-2007, 02:16 PM

adjust yards per attempt

lastchance · #3 10-01-2007, 03:03 PM

[ QUOTE ]
adjust yards per attempt

[/ QUOTE ]

jogsxyz · #4 10-01-2007, 03:45 PM

[ QUOTE ]
adjust yards per attempt

[/ QUOTE ]

Which should include sacks/rushing yards. Now we only
need to know how to adjust for TDs and INTs.

Jack of Arcades · #5 10-01-2007, 03:49 PM

Pro Football Reference/Football Guys adds 10 yards for a td and subtracts 45 for an INT.

THAY3R · #6 10-01-2007, 02:16 PM

It's funny how media people use Passer rating like it's gospel yet freak out when someone mentions OBP.

kyleb · #7 10-01-2007, 02:57 PM

Much less VORP.

PokerFink · #8 10-01-2007, 03:08 PM

kevkev,

Could you post some 2006 quarterback statistics to see how your rating compares with the normal rating? Don't need all their stats obv, just conventional QB rating and your rating.

Let's pick 10 QBs to compare: Brady, Manning, Palmer, Leftwich, VY, Grossman, David Carr, JP Losman, Jeff Garcia and Roethlisberzerswersr.

Edit: I'm not trying to be an ass here, I'm legit curious, but I'm on my parents computer and they don't have Excel. And those QBs aren't random, some of those names are in there for specific reasons.

kevkev60614 · #9 10-01-2007, 03:45 PM

[ QUOTE ]
kevkev,

Could you post some 2006 quarterback statistics to see how your rating compares with the normal rating? Don't need all their stats obv, just conventional QB rating and your rating.

Let's pick 10 QBs to compare: Brady, Manning, Palmer, Leftwich, VY, Grossman, David Carr, JP Losman, Jeff Garcia and Roethlisberzerswersr.

Edit: I'm not trying to be an ass here, I'm legit curious, but I'm on my parents computer and they don't have Excel. And those QBs aren't random, some of those names are in there for specific reasons.

[/ QUOTE ]
Name Current Proposed
Brady 87.9 57.5
Manning 101.0 70.7
Palmer 93.9 66.0
Leftwich 79.0 50.0
VY 66.7 44.2
Grossman 73.9 49.7
David Carr 82.1 46.3
JP Losman 84.9 55.8
Jeff Garcia 95.8 64.6
Roethlisberger 75.4 51.5

Interesting: They're much more lumped around average than I expected. I'd been looking at the stats on a game-by-game basis, not by a season. Probably expected because the proposed range is 2/3 the size of the current range. Maybe that's bad?

Also, by both measures Leftwich was pretty much exactly average. David Carr is slightly above average currently but slightly below according to the proposed measure. Probably due to a relatively high completion percentage (68.3%).

kevkev60614 · #10 10-01-2007, 03:55 PM

Current Proposed Adj YPA
Brady 87.9 57.5 6.26
Manning 101 70.7 7.72
Palmer 93.9 66.0 7.17
Leftwich 79 50.0 5.49
VY 66.7 44.2 4.86
Grossman 73.9 49.7 5.26
David Carr 82.1 46.3 5.29
JP Losman 84.9 55.8 6.09
Jeff Garcia 95.8 64.6 7.02
Roethlisberger 75.4 51.5 5.67

I added Adj YPA for completion's sake. I hadn't ever seen it before but my measure was surprisingly similar. Adj YPA has a lower correlation (based on '06 data only) to victory, and the difference in opposing teams Adj YPA's has a lower correlation to victory, but it's pretty marginal. I don't even know whether those things matter. Meh.

It's also nice that Adj YPA is not capped at all (like passer rating can be capped in four spots and my made-up rating can be capped by 0 or 100). I still kind of feel that a number based on a range from 0-100 makes more intuitive sense to me, but Adj YPA is definitely a simpler number to calculate than passer rating. Meh.

EDIT: The only diff between the top 10 rankings based on my methodology vs Adj YPA is between Grossman and Carr. Who do you think had the better year?

Name Comp Att Pct Y/A Yds TD INT
Grossman 262 480 54.6 6.65 3193 23 20
Carr 302 442 68.3 6.26 2767 11 12