PDA

View Full Version : More QB rating bashing



Andy Freeland
07-30-2010, 03:04 PM
This one is better. I took every QB that qualified over the last 3 years. They are sorted by rating (highest on the left to lowest on the right). Then I broke out each individual component, graphed against the raw rating /4. You can clearly see that INT and comp Pct is a much bigger component, just like Kiran said. yards/attempt and especially TD pct are much, much less significant.
124

Let me know if that explanation made any sense.

Andy Freeland
07-30-2010, 03:07 PM
125
click on image for full sized picture

Phil McConkey
07-30-2010, 04:02 PM
That is exactly why the QB's of the past have lower ratings. Everything back then was 7 sep drop which made for better yds/attempt but lower completion percentage. Back then, if you completed better than 50% of your passes, you were doing good. The west coast offense and it's reliance on 3 and 5 step drops helped the completion % rise and along with it, the QB ratings.

ScottDCP
07-30-2010, 05:31 PM
This one is better. I took every QB that qualified over the last 3 years. They are sorted by rating (highest on the left to lowest on the right). Then I broke out each individual component, graphed against the raw rating /4. You can clearly see that INT and comp Pct is a much bigger component, just like Kiran said. yards/attempt and especially TD pct are much, much less significant.
124

Let me know if that explanation made any sense.

Just to be clear:

This supports the assertion that the current QB rating system rewards lack of errors much more than it rewards tremendous success. Which supports most everybody's gut feeling that the rating system is hinky.

Andy Freeland
07-30-2010, 06:06 PM
Just to be clear:

This supports the assertion that the current QB rating system rewards lack of errors much more than it rewards tremendous success. Which supports most everybody's gut feeling that the rating system is hinky.

That is correct. Over the last 3 years there have been 98 QBs qualify for the passing title, only 1 had a higher score in the TD component of the rating than the INT component (Sage Rosenfels in '07: 15 TDs - 12 INTs in 240 attempts) and none had a higher score in yards/attempt than in completion pct.

msclemons
07-30-2010, 11:40 PM
That's pretty amazing consistency there. Int/Att and cmp% seem to be locked in at .5 above TD/Att and yds/Att. Except that one guy in the middle of the graph with the big spike on Int/Att. Out of curiosity, who was that? Cutler?

Do you think the consistency might indicate that - while the "safety" stats are overvalued - the system as a whole works? TD/Att and yds/Att seem to follow the other two pretty closely, albeit .5 points lower.

Andy Freeland
07-30-2010, 11:53 PM
That's pretty amazing consistency there. Int/Att and cmp% seem to be locked in at .5 above TD/Att and yds/Att. Except that one guy in the middle of the graph with the big spike on Int/Att. Out of curiosity, who was that? Cutler?

Do you think the consistency might indicate that - while the "safety" stats are overvalued - the system as a whole works? TD/Att and yds/Att seem to follow the other two pretty closely, albeit .5 points lower.

I wondered the same thing about Cutler, it was actually Trent Edwards.

I see what you're saying about the system working and I wondered the same thing, but after I thought about it that's really not the case. It does value not throwing INTs over throwing TDs. Think about if (I should probably say when) INT/Comp% was 10 times greater than TD/Yards, then TDs would have little or no value. I know that's an extreme example, but it does show that the discrepancy does affect the final rankings.

I'll be doing a full post on this later, but after doing some more research I realized that this discrepancy didn't exist when the rating was created in 1971. All 4 graphs were similar and rating went right down the middle of them. Meaning that the rating was written specifically for the stats of the day and is outdated now.

msclemons
07-31-2010, 12:09 AM
The graph is a great reflection of the direction the league has gone. The short, high-percentage, low risk passing game seems to be almost universal now.

The cmp% and yds/Att lines really show that - they're almost identical. Which means everyone is attempting throws for the same distance; cmp% and yds/att almost have a fixed ratio. (I know I phrased that wrong - it's been 20 years since my last algebra class).

Andy Freeland
07-31-2010, 12:15 AM
it's been 20 years since my last algebra class.

Which was 5 years after mine.

msclemons
07-31-2010, 12:17 AM
I was trying to seem younger and more hip. You're the math wizard, is ratio the correct term there?

Andy Freeland
07-31-2010, 12:23 AM
I think ratio is right. If the ratio (I'll use it too, to make you feel better) of yards/att to comp/att is roughly the same league-wide then that would mean that all teams are throwing roughly the same distance on most passes.