Originally posted by BobS:
I am waiting for an Alex Smith fan to come up with a system that makes 2-7 for 10 yards look good.

LOL dude c'mon. Find us ONE person who is defending "2/7 for 10 yards."

s**t like this is so irritating in Niner Talk. Someone makes ONE reasonable Alex Smith post (i.e. "He wasn't good but this is why and he needs to play better...it's hist last chance.") And some how that equates to: "I LUV TEH SMYTH AND DEEFEND TIL DETH DOO US PART DOOD!"

Originally posted by kray28:
Measures of "clutchness" btw....highly problematic. There is no rigorous way to define "clutch". It's a fundamentally qualitative interpretation trying to masquerade as an objective metric.

The fact that such a "measure" is in the formula for this QBR rating basically invalidates it.

how do you know? they can have a statistic that shows that with 7+ points deficit with one quarter or less to go, what % of the time do they win, and what is the average # of points are they able to score. This is highly calculable.

Originally posted by kray28:
Basically they designed a rating system to fit their preconceived notion of what the ratings should look like. Basically a statistical version of "begging the question". Another good example of this is John Hollinger's PER system for basketball, basically weighted to assume that Michael Jordan was the greatest player to ever live, and weighting his particular statistical strengths as the most important criteria in rating everyone else.

This. My problem with the new TQR is its reliance on a lot of subjective factors. Therefore when the judge(s) gave percentages to certain categories, he/they would unconsciously bias towards established, more visible QBs. Not that the ratings would change significantly for the top players, but the guys playing for worse teams (Schaub for example) tend to be at a disadvantage.

Originally posted by Shaj:
I doubt it is BS. I think what they probably did is some very intense statistics involving correlation coefficients and determining what QB factors most influenced winning games and from there assigned variables and weights. The previous system was somewhat arbitrary and missed a lot of variables. I'm sure they then took this model and ran the figures for previous years to see if in fact the results supported empirical results on the field and W/L ratios. I'm sure the analysis was very much like what the A's did with Moneyball. Correlation coefficient analysis is the key here.

It also makes sense in that before, Alex was being compared to QBs he had no business being compared to, such as Carson Palmer, and now that comparison is no longer valid with this system. It however introduces some more interesting comparisons, such as why Bradford, who seemed very clutch, is so close to Smith, who is anything but. However, Bradford is just a rookie, and Smith of course is a 7 year vet.

What I said had nothing to do with Smith. But you didn't comment on my Matt Schaub comment.

Originally posted by Shaj:
how do you know? they can have a statistic that shows that with 7+ points deficit with one quarter or less to go, what % of the time do they win, and what is the average # of points are they able to score. This is highly calculable.

How would they do it exactly? What if it was a special team or defensive play that accounted for the points? What if a running back breaks one 90 yards? What if the other team has a turnover in the red zone and after a couple of dive plays the game winning field goal is kicked? Does the QB get credit for everything or does he have to be involved in how the points are scored? Does he have to pass for X amount of yards or is handing the ball off 15 times to Jerome Bettis good enough? It seems fairly problematic to me.

how do you know? they can have a statistic that shows that with 7+ points deficit with one quarter or less to go, what % of the time do they win, and what is the average # of points are they able to score. This is highly calculable.

How would they do it exactly? What if it was a special team or defensive play that accounted for the points? What if a running back breaks one 90 yards? What if the other team has a turnover in the red zone and after a couple of dive plays the game winning field goal is kicked? Does the QB get credit for everything or does he have to be involved in how the points are scored? Does he have to pass for X amount of yards or is handing the ball off 15 times to Jerome Bettis good enough? It seems fairly problematic to me.

Originally posted by kray28:
Measures of "clutchness" btw....highly problematic. There is no rigorous way to define "clutch". It's a fundamentally qualitative interpretation trying to masquerade as an objective metric.

The fact that such a "measure" is in the formula for this QBR rating basically invalidates it.

Of course you can define "clutch". Whether or not you agree with the criteria they use is another issue. I think they probably used something like 4th quarter stats with the game being within 10 pts or something along those lines. I have no problem with that.

Originally posted by kray28:
Basically they designed a rating system to fit their preconceived notion of what the ratings should look like. Basically a statistical version of "begging the question". Another good example of this is John Hollinger's PER system for basketball, basically weighted to assume that Michael Jordan was the greatest player to ever live, and weighting his particular statistical strengths as the most important criteria in rating everyone else.

Your argument would make sense if the top QBs on that list weren't objectively...the top QBs in the NFL. I don't think you'll find too many oddities in the way the QBs are ranked based on Total QBR system.

People perceive that Tom Brady and Peyton Manning are the best QBs in the NFL because...they ARE the best QBs in the NFL.