Alex Smith QBR: Comparing Week 5 and Week 9

what i don't like about the QBR is that it harshly punishes sacks and greatly rewards early TD's, but not late ones. Most people would agree that this years Philly game was one of Alex's best games in his career, if not the best, and his rating for that game was abysmal.

Having a perfect 3rd quarter on the road in a game that your losing is exactly what great players do, and what makes the QB position important. If the QBR can't measure that then they need to go back to the drawing board.

Originally posted by LieutKaffee:

Originally posted by miked1978:
Now I don't know a lot about the QBR rating but doesn't it take into account 3rd down conversions? I beleive the Niners were up to their old tricks on 3rd down converting will below 20%

We were 3-12 against the Redskins and 2-7 against the Bucs. By percentage they are almost equally bad, but I couldn't find out more just by looking at box scores. It's possible more of our 3rd downs this week "mattered" since the game was more competitive longer. It's also possible Alex failed on more 3rd downs he should have had. For example I'm sure missing on a 3rd and 3 pass is worse for QBR than missing on 3rd and 11.

An attempt to pick up the 3rd down by running should not be reflected in QBR. The only justification would be that the QB COULD audible to a pass but that assumes the package in the game at the time would allow it.

Trying to make too much out of it...

Mike Sando did address this after the Tampa Bay game (he talks about it each week). In the Tampa game, Alex played great in the first half when the outcome of the game was in doubt. In the second half, he player poorly, but it didn't matter because it didn't have an impact on the outcome of the game.

I do think that Alex shows the flaws in QBR. He is probably not as good as his passer rating, but neither is he as bad as his QBR either. It is kind of a shame, because you would think that with all the effort into analysis to produce the QBR, it would be better than the passer rating, but Alex shows that it also makes mistakes as well.

Originally posted by dj43:
Ironic that pro sports try so hard to control gambling by players/coaches but the entire injury reporting system is set up to PROMOTE gambling interests. Reporting injuries has ZERO impact on how the game is actually played. The only logical reason to report on Wednesday is to give the bookies more time to set the lines and promote their business.

I doubt injury reports influence betting lines much at all. There are no penalties for playing someone listed probable/questionable/doubtful. If the injury reports had any binding power, they would influence the lines. As the system is set up right now it's really a waste of time.

Plus early lines are released for the week's slate before injury reports are published.

But...if you're saying that the injury reports are a way of generating betting interest/action from the public, you may have a point.

Oops wrong thread

I was eager to read the QBR after the Philly game, it was abysmal, and Sando's explanation was stupid. According to him Alex did nothing to justify a better QBR becuase when he threw the first TD we simply cut the deficit do 13 points, etc.

When you read the explantion from the Tampa game it's the opposite, he threw the TDs early with the game on the line.

WTF???

Originally posted by JeuSF49:
I was eager to read the QBR after the Philly game, it was abysmal, and Sando's explanation was stupid. According to him Alex did nothing to justify a better QBR becuase when he threw the first TD we simply cut the deficit do 13 points, etc.

When you read the explantion from the Tampa game it's the opposite, he threw the TDs early with the game on the line.

WTF???

Well, stats are all about...statistics.

Niners down 23-3 at Philly have maybe a 2% chance of winning (statistically). Smith throws a TD, now we're down 23-10..maybe a 7% chance now. Even the second TD pass, to make it 23-17...still only about a 20% chance. So, two nice drives, two nice results...but only about a 20% increase in win probability. (The last TD drive to take the lead was almost all run plays, so Smith got nada for it.)

Whereas the Tampa games, his first TD pass was probably worth 20% by itself.

Stats will always struggle with context. It's hard to separate "garbage time" from "start of epic comeback time." John Beck drove the skins down for a late TD that cut it to a one score game. Was that garbage time, or the start of an epic comeback (that didn't happen only because the onside kick failed)?

Personally I don't have a problem with this part of QBR. Statistically speaking, some TDs are worth far more than others and should be credited as such.

is it possible in X amount of years that the NFL will drop it's current rating format and adopt this ESPN QBR b******t?

Originally posted by johnnyredneat:
Fellow stat geek here too.

I think the answer the OP's question (what is QBR measuring that made the #s so different in the TB and Wash games) is that QBR focuses more on "win probability added", not TD themselves. Basically, every point in a football game can be shown as an expected point value. If SF begins a drive at the 20, then that drive might be expected to produce 1 point for SF (on average). If Alex throws a 20 yard pass, now it's 1st and 10 on the 40, and the drive can now be expected to produce 2 points on average, and so on. So that throw by Alex was "worth" 1 point.

Without getting too deep into the numbers then I think the QBR difference in the Wash and TB games comes down to this: in the TB game (especially early, when the game was in doubt) we didn't have great field position to start. Alex had to drive a long-ish field to score. And we did it mostly by passing – as I remember, we surprised everyone by throwing a lot to open the game. So Alex got a lot credit not just for the Delanie TD but also for the long drives that preceded it. Basically, in QBR's eyes we started a drive with a low expected point value, and Alex turned it into 7 points [almost] all on his own.

In the Wash game, we consistently had excellent field position (i.e. a higher initial point value for the drive) but didn't convert it into lots of points, so Alex got penalized for it. On the Miller TD, we started practically in the red zone after the fumble, so the drive would be "expected" to produce almost a TD. So Alex's throw was just an "average" play in QBR's eyes.

Many things wrong with QBR in my eyes. But this might be at least an explanation.

Very good insight. Thanks for the input!

Originally posted by dj43:

Originally posted by LieutKaffee:

Originally posted by miked1978:
Now I don't know a lot about the QBR rating but doesn't it take into account 3rd down conversions? I beleive the Niners were up to their old tricks on 3rd down converting will below 20%

We were 3-12 against the Redskins and 2-7 against the Bucs. By percentage they are almost equally bad, but I couldn't find out more just by looking at box scores. It's possible more of our 3rd downs this week "mattered" since the game was more competitive longer. It's also possible Alex failed on more 3rd downs he should have had. For example I'm sure missing on a 3rd and 3 pass is worse for QBR than missing on 3rd and 11.

An attempt to pick up the 3rd down by running should not be reflected in QBR. The only justification would be that the QB COULD audible to a pass but that assumes the package in the game at the time would allow it.

Trying to make too much out of it...

Sorry for the confusion. I was referring to 3rd downs in which we passed it, which I assumed was probably a good proportion of them.

Originally posted by johnnyredneat:

Originally posted by JeuSF49:
I was eager to read the QBR after the Philly game, it was abysmal, and Sando's explanation was stupid. According to him Alex did nothing to justify a better QBR becuase when he threw the first TD we simply cut the deficit do 13 points, etc.

When you read the explantion from the Tampa game it's the opposite, he threw the TDs early with the game on the line.

WTF???

Well, stats are all about...statistics.

Niners down 23-3 at Philly have maybe a 2% chance of winning (statistically). Smith throws a TD, now we're down 23-10..maybe a 7% chance now. Even the second TD pass, to make it 23-17...still only about a 20% chance. So, two nice drives, two nice results...but only about a 20% increase in win probability. (The last TD drive to take the lead was almost all run plays, so Smith got nada for it.)

Whereas the Tampa games, his first TD pass was probably worth 20% by itself.

Stats will always struggle with context. It's hard to separate "garbage time" from "start of epic comeback time." John Beck drove the skins down for a late TD that cut it to a one score game. Was that garbage time, or the start of an epic comeback (that didn't happen only because the onside kick failed)?

Personally I don't have a problem with this part of QBR. Statistically speaking, some TDs are worth far more than others and should be credited as such.

Another great explanation. I think your post here does point to something QBR should address, because I agree with the above poster that putting together a perfect 3rd quarter on the road while down 20 points is, in fact, the mark of a great QB.

Your point about Beck was well taken though. You can only tell in hindsight whether something was garbage time or the start of an epic comeback. Perhaps QBR should have a mechanism to weight the 3rd quarter more cumulatively after the comeback was complete.

Originally posted by Paul:
is it possible in X amount of years that the NFL will drop it's current rating format and adopt this ESPN QBR b******t?

Passer Rating came around in the 1970s or something and pretty much no one speaks positively about it. So to an extent, yeah, I assume these new stats are attempts to ultimately replace it as the number 1 tool.

On the other hand though, if you sort the current season by Passer Rating, you do tend to get a pretty good picture of who the best QBs are. At that rudimentary level (simply comparing the top 5-10 guys to an eye test), it may still fare better than QBR at the moment, which is ironic.

I like the Passer Rating system because it more concretely represents that actual stats of a QB rather than the "effectiveness" of that QB. When I see a Passer Rating of 100, I know that if I look at that QB's stats, I will see some pretty numbers.

The QBR, not so much. It's too subjective and you can't decipher very much by looking at the QBR alone. That is where this measure fails. The entire purpose of a stat is the fact that you can infer something based on that stat. One can't really infer very much based on the QBR. In fact, more questions arise from the QBR such as, "he threw 3 TD's with 0INTs, so why is the rating so low. One is required to read the explanation report of the player's QBR. There's so many parts that can lower the QBR that it's not visible from the stat alone.

Therefore, the QBR fails as a stat because it requires a full report on what factors lowered that players QBR from another player who had identical hard stats but a higher QBR whereas the Passer Rating can be taken as a hard stat and the overall stats of the QB can be inferred fairly accurately.

Originally posted by LieutKaffee:
Passer Rating came around in the 1970s or something and pretty much no one speaks positively about it. So to an extent, yeah, I assume these new stats are attempts to ultimately replace it as the number 1 tool.

On the other hand though, if you sort the current season by Passer Rating, you do tend to get a pretty good picture of who the best QBs are. At that rudimentary level (simply comparing the top 5-10 guys to an eye test), it may still fare better than QBR at the moment, which is ironic.

No matter the issues with QBR, passer rating is a worse stat. I can't see anyone arguing otherwise. Any stat that counts a clock-stopping spike the same as any other miss, or a hail-mary INT the same as any other pick, is pretty awful.

Most any stat is going to be good at the margins. Anything, even something as bad as passer rating, can show you who's really really good and who's really really bad. But it's the middle third that reveals the most.

Originally posted by johnnyredneat:

Originally posted by LieutKaffee:
Passer Rating came around in the 1970s or something and pretty much no one speaks positively about it. So to an extent, yeah, I assume these new stats are attempts to ultimately replace it as the number 1 tool.

On the other hand though, if you sort the current season by Passer Rating, you do tend to get a pretty good picture of who the best QBs are. At that rudimentary level (simply comparing the top 5-10 guys to an eye test), it may still fare better than QBR at the moment, which is ironic.

No matter the issues with QBR, passer rating is a worse stat. I can't see anyone arguing otherwise. Any stat that counts a clock-stopping spike the same as any other miss, or a hail-mary INT the same as any other pick, is pretty awful.

Most any stat is going to be good at the margins. Anything, even something as bad as passer rating, can show you who's really really good and who's really really bad. But it's the middle third that reveals the most.

Right. So I assume you would have answered similarly, that something like QBR may eventually displace Passer Rating as the most used stat.