Smidgeon
06-21-2011, 06:26 PM
Another statistics heavy post.
So one of the shortcomings of the passer rating formula (in my mind) is that it doesn't take into account how successful a QB is in maintaining drives. For example, a sack can kill a drive and a first down continues it.
So based on the original passing formula, I added two more components. One, the sack percentage and one the first down percentage.
The passer rating's formula is designed to take into account several factors within certain boundaries so that there's a clear minimum and a clear maximum. It's weighted so that each of the components gets a fair portion of the rating.
The first component is completion percentage:
(Comp/Att-.3)x5
The second component is yards per attempt:
(Yards/Att-3)x.25
The third component is Touchdown percentage:
(TD/Att)x20
The fourth component is Interception percentage:
2.375-((Int/Att)x250
Then you add all those numbers up, divide by 6 and then multiply by 100.
Don't know why they do it like that, but it's the established method. So I used that framework to add two more pieces:
The fifth component is sack percentage:
2.37((Sack/Att)x5)
The sixth component is First down rate:
(First downs/Att+0.6)*5-3
I then added all six components together, divided by 8, and then multiplied by 100.
The results of the starters?
In 2009:
* Rodgers (with his 50 sacks) dropped 2 spots in the passer rankings, ending up at 6 overall. It was the second largest drop in rankings in the league.
* Matt Cassel dropped 3 spots, the most in the league.
* The largest jump was by Sanchez, moving up 3 spots to 25.
* Vince Young moved up 2 spots based on his ability to avoid sacks.
* How bad was JaMarcus Russell? He ranked last of all starters in 4 of the 6 categories (31st in yards/attempt--Quinn and 27th in Int %--Freeman).
* Rodgers ranked 29th in sack percentage, despite tying for dead last in number of sacks with Roethlisberger. The others with worse percentages were Ryan Fitzpatrick (21 sacks in 227 attempts) Roethlisberger (50 in 506) and Russell (33 in 246).
In 2010 (only 31 calculated starters due to Cleveland's musical chairs):
* The top three positions and the bottom four didn't change order at all based on the rankings calculation change. It was still Brady/Rivers/Rodgers at the top and Hasselbeck/Favre/Anderson/Clausen at the bottom.
* Clausen in 2010 wasn't as bad as Russell in 2009. He was deadlast in only three categories (yards/att, TD/att, first down/att), with Anderson last in completion percentage (Clausen at 30th), Farve in interception percentage (Clausen at 19th), and Cutler in sack percentage (Clausen at 30th).
* Peyton Manning jumped 4 spots (the biggest jump in either year).
* Flacco lost 5 positions and Campbell lost 4. Not a lot of redemption for them, which means their sack ratings and first down ratings were average to poor.
* Vick lost 1 spot. Compared to Young, I think one of two conclusions can be jumped to. Of the two mobile QBs, either one is naturally that much better at avoiding sacks (Young), or Vick truly is a passing QB and isn't that good at pocket presence yet--but which proves he's not looking to run at first fright.
* Rivers' first down percentage was 4.5% higher than #2, and #2 through #5 in that category were all within 0.4% of each other.
Overall thoughts:
* Peyton Manning is clearly the king of sack percentage. For both 2009 and 2010, he was best in that category. Drew Brees was the only other one to be in the Top 5 in both years and Matt Ryan and Tom Brady in the Top 10.
* Rivers was the best of 1st down % (too bad it doesn't win playoff games) in both years. Although many of the same QBs were in the Top 5 and 10 in this category.
* Adding these two components moves to some shifting of the rankings. The best QBs still end up at the top and the worst at the bottom. Most of the time, the shifts are small, but I think these components reward the QBs who make smart decisions. Just thoughts.
So what does this all mean? Nothing. Nada. It's just fun. And I may still have to tweak the last two components just to make sure their minimums and maximums fit the philosophy of the other components.
So one of the shortcomings of the passer rating formula (in my mind) is that it doesn't take into account how successful a QB is in maintaining drives. For example, a sack can kill a drive and a first down continues it.
So based on the original passing formula, I added two more components. One, the sack percentage and one the first down percentage.
The passer rating's formula is designed to take into account several factors within certain boundaries so that there's a clear minimum and a clear maximum. It's weighted so that each of the components gets a fair portion of the rating.
The first component is completion percentage:
(Comp/Att-.3)x5
The second component is yards per attempt:
(Yards/Att-3)x.25
The third component is Touchdown percentage:
(TD/Att)x20
The fourth component is Interception percentage:
2.375-((Int/Att)x250
Then you add all those numbers up, divide by 6 and then multiply by 100.
Don't know why they do it like that, but it's the established method. So I used that framework to add two more pieces:
The fifth component is sack percentage:
2.37((Sack/Att)x5)
The sixth component is First down rate:
(First downs/Att+0.6)*5-3
I then added all six components together, divided by 8, and then multiplied by 100.
The results of the starters?
In 2009:
* Rodgers (with his 50 sacks) dropped 2 spots in the passer rankings, ending up at 6 overall. It was the second largest drop in rankings in the league.
* Matt Cassel dropped 3 spots, the most in the league.
* The largest jump was by Sanchez, moving up 3 spots to 25.
* Vince Young moved up 2 spots based on his ability to avoid sacks.
* How bad was JaMarcus Russell? He ranked last of all starters in 4 of the 6 categories (31st in yards/attempt--Quinn and 27th in Int %--Freeman).
* Rodgers ranked 29th in sack percentage, despite tying for dead last in number of sacks with Roethlisberger. The others with worse percentages were Ryan Fitzpatrick (21 sacks in 227 attempts) Roethlisberger (50 in 506) and Russell (33 in 246).
In 2010 (only 31 calculated starters due to Cleveland's musical chairs):
* The top three positions and the bottom four didn't change order at all based on the rankings calculation change. It was still Brady/Rivers/Rodgers at the top and Hasselbeck/Favre/Anderson/Clausen at the bottom.
* Clausen in 2010 wasn't as bad as Russell in 2009. He was deadlast in only three categories (yards/att, TD/att, first down/att), with Anderson last in completion percentage (Clausen at 30th), Farve in interception percentage (Clausen at 19th), and Cutler in sack percentage (Clausen at 30th).
* Peyton Manning jumped 4 spots (the biggest jump in either year).
* Flacco lost 5 positions and Campbell lost 4. Not a lot of redemption for them, which means their sack ratings and first down ratings were average to poor.
* Vick lost 1 spot. Compared to Young, I think one of two conclusions can be jumped to. Of the two mobile QBs, either one is naturally that much better at avoiding sacks (Young), or Vick truly is a passing QB and isn't that good at pocket presence yet--but which proves he's not looking to run at first fright.
* Rivers' first down percentage was 4.5% higher than #2, and #2 through #5 in that category were all within 0.4% of each other.
Overall thoughts:
* Peyton Manning is clearly the king of sack percentage. For both 2009 and 2010, he was best in that category. Drew Brees was the only other one to be in the Top 5 in both years and Matt Ryan and Tom Brady in the Top 10.
* Rivers was the best of 1st down % (too bad it doesn't win playoff games) in both years. Although many of the same QBs were in the Top 5 and 10 in this category.
* Adding these two components moves to some shifting of the rankings. The best QBs still end up at the top and the worst at the bottom. Most of the time, the shifts are small, but I think these components reward the QBs who make smart decisions. Just thoughts.
So what does this all mean? Nothing. Nada. It's just fun. And I may still have to tweak the last two components just to make sure their minimums and maximums fit the philosophy of the other components.