Welcome, Guest. Please login or register.
January 15, 2025, 10:48:04 am
Home Help Search Calendar Login Register
News: Brian Fein is now blogging weekly!  Make sure to check the homepage for his latest editorial.
+  The Dolphins Make Me Cry.com - Forums
|-+  TDMMC Forums
| |-+  Dolphins Discussion (Moderators: CF DolFan, MaineDolFan)
| | |-+  I'm warming to the idea of a re-tread QB.
« previous next »
Pages: 1 2 3 [4] 5 Print
Author Topic: I'm warming to the idea of a re-tread QB.  (Read 16322 times)
Brian Fein
Global Moderator
Uber Member
*****
Posts: 28291

WHAAAAA???

chunkyb
« Reply #45 on: January 19, 2011, 09:37:31 am »

Any stat in which the first term is "adjusted" is blatantly made-up crap, IMO.
Logged
Spider-Dan
Global Moderator
Uber Member
*****
Posts: 15936


Bay Area Niner-Hater


« Reply #46 on: January 19, 2011, 12:26:24 pm »

In terms of pure statistical value (i.e. contributing to a win), interceptions are being given too negative a weight. And I don't much like placing sack yards on the quarterback. At least not with that high a weight (same as passing yards). That might be their statistical contribution to the game outcome, but as a tool for quarterback evaluation, it seems like there's little to favor ANY/A over standard passer rating.  That might be their statistical contribution to the game outcome, but as a tool for quarterback evaluation, it seems like there's little to favor ANY/A over standard passer rating.
Without even looking into the statistical breakdown, I think it's perfectly reasonable to weigh interceptions at -45.  An interception is both the lack of opportunity to further advance the ball on subsequent downs in that set, and a lost opportunity to punt.  Do you think that it's unreasonable to say that an interception is, on average, 45 yards of lost field position, compared to a drive in which an INT is not thrown?

As for other yardage valuations (and the ANY/A calculation in general), I'll just say this: the numbers that they are using in that formula are far from arbitrary.  They put a lot of work into determining those values (as you can see from the discussion of how they arrived at a TD being worth 20 yards).

Now, if you want to argue that stats are overrated and you can't quantify heart (or something similar), then fine.  But if you are going to enter the realm of statistical comparison, ANY/A has a pretty solid foundation.

Furthermore, if you are going to choose a metric to compare to, passer rating is a bad one; it greatly overvalues completion percentage.  There is an oft-cited example of passer rating's shortcomings: the QB who goes 3/3 for 9 total yards will have a better rating than the QB who goes 1/3 for 30 yards.
Logged

fyo
Uber Member
*****
Posts: 7545


4866.5 miles from Dolphin Stadium


« Reply #47 on: January 19, 2011, 01:54:39 pm »

Now, if you want to argue that stats are overrated and you can't quantify heart (or something similar), then fine.  But if you are going to enter the realm of statistical comparison, ANY/A has a pretty solid foundation.

I'll start with the last.

The problem is not statistics, it's statistics that lack context. Attempting to improve upon passer rating at a comparison metric without taking ANY context into consideration is ridiculous, IMHO.

Quote
Without even looking into the statistical breakdown, I think it's perfectly reasonable to weigh interceptions at -45.  An interception is both the lack of opportunity to further advance the ball on subsequent downs in that set, and a lost opportunity to punt.  Do you think that it's unreasonable to say that an interception is, on average, 45 yards of lost field position, compared to a drive in which an INT is not thrown?

I strongly dislike the conversion of everything to yards. While perhaps somewhat arbitrary, I'd prefer if the conversion were to points instead. It's much easier to argue how many points a given number of yards is worth than what number of yards a touchdown should be converted to. Or, even better, just call it a "score" and don't pretend its yards or points.

I've seen statistical analysis that puts the "cost" of an interception at around 3.5 points, or half a touchdown. Now, if you look at the average starting point in the NFL and without looking anything up, I'd guesstimate around the 30 yard line. That puts the value of a touchdown at roughly 70 yards and the value of an interception at half that, i.e. 35 yards.

Now, if you dole out points for yards alone, regardless of what happens on those drives, things are clearly being counted double. The 70 yards you moved and the 7 points you got. How you decide to split the difference is somewhat arbitrary, but key when comparing to interceptions. Splitting the points equally, you could put the value of a touchdown at 3.5 points or 35 yards.

That would mean 35 and -35 in the ANY/A calculation instead of the 20, -45 factors used. I.e. the "cost" of an interception compared to a touchdown is being overvalued by more than a factor of 2.

Quote
There is an oft-cited example of passer rating's shortcomings: the QB who goes 3/3 for 9 total yards will have a better rating than the QB who goes 1/3 for 30 yards.

Like I said, any metric that doesn't take context into considering AT ALL is doomed to fail. It's easy to pick and choose scenarios where one or the other looks ridiculous. Completion percentage has a big value when there's a lot of passes (how relevant is it that the metric work in 3-pass situations?) since it means being able to churn out long drives, which not only scores points but also allows running out the clock. ANY/A completely ignores this aspect.

If you're going to replace passer rating by a "simple" compound metric, why not include downs? A first down in and of itself has value. There's also a big difference in "points lost" on an interception thrown on first down and on third down. This would be simple enough to incorporate in the simple compound metric. (And by simple, I'm referring to a lack of regression analysis, not taking time remaining, position on field, current score, or opponent strength into account).
Logged
JVides
Uber Member
*****
Posts: 2915



« Reply #48 on: January 19, 2011, 02:07:18 pm »

I'd like to say that the accountant, who lives in a world filled with statistics, metrics, and other measurement tools finds this discussion of QB rating metrics extremely boring.  If I have to be a stats geek to move past "casual fan" status, as one of you suggests, then casual fan I am.

Unless this adjusted formula changes with the times to account for changes made to the game, it's a semi-useless tool as well (How does Otto Graham fare by this metric, or even Sammy Baugh?).  How does the physical corner play permitted during the 70s and 80s (or worse, the early years of the sport) compare to the "don't touch 'em" policies of the current era?
Logged

"under wandering stars I've grown
by myself but not alone
I ask no one"
Metallica, "Wherever I may Roam"
Spider-Dan
Global Moderator
Uber Member
*****
Posts: 15936


Bay Area Niner-Hater


« Reply #49 on: January 19, 2011, 02:51:17 pm »

The problem is not statistics, it's statistics that lack context. Attempting to improve upon passer rating at a comparison metric without taking ANY context into consideration is ridiculous, IMHO.
What kind of context are you talking about?  Do you require the same context when you say that Chris Johnson had a 2000-yard season, or that Jim Brown averaged 5.2 yards per carry, or that Peyton Manning's career passer rating is 94.9?

Quote
I strongly dislike the conversion of everything to yards. While perhaps somewhat arbitrary, I'd prefer if the conversion were to points instead.
Then you should be happy to learn that they have also provided a conversion of yards to points (and vice versa).

From the article I linked, an "average passing yard" is worth .0653 points, and one point is worth 15.3 "average passing yards".

Quote
I've seen statistical analysis that puts the "cost" of an interception at around 3.5 points, or half a touchdown.
3.5 * 15.3 = 53.55 yards

That would seem to place an even bigger value on interceptions than the 45-yard number that you complained about.

Quote
Now, if you look at the average starting point in the NFL and without looking anything up, I'd guesstimate around the 30 yard line. That puts the value of a touchdown at roughly 70 yards and the value of an interception at half that, i.e. 35 yards.
I'm not going to get too far into the numbers game here.  I know enough about the statistical crunching to realize that we are both FAR out of our depth; while I don't pretend to know all the variables, I've read enough articles on the subject to recognize your evaluation as far too simplistic to be considered comparable.  Again, I suggest you read the article I linked in my previous post.

Quote
If you're going to replace passer rating by a "simple" compound metric, why not include downs?
What makes you think first downs aren't part of the metric?

The ANY/A is a simplified metric, used to calculate the value of passing yards per attempt without having to manually break down each attempt by individual situation.  It is the result of combining many smaller, more situation-specific formulas.  Now, if you prefer to go through the game logs and break down each individual pass, knock yourself out; there is a plethora of statistical formulas available that will allow you to break down the effectiveness of each individual pass in context, giving you the most accurate evaluation possible.  But if you want a formula that can be applied to every QB fairly easily, and still give more contextual value than simple passer rating, ANY/A is the way to go.
« Last Edit: January 19, 2011, 02:52:57 pm by Spider-Dan » Logged

Pappy13
Uber Member
*****
Posts: 8359



« Reply #50 on: January 19, 2011, 03:44:22 pm »

The problem that I have with stats like this one is not the formula itself, it's the rankings of the players using that formula and what we read into them.  Ok, we put all this stuff into a formula and pull out this stat that shows that a QB who had a GREAT year had an 8.2 ANY/A this year, but an average one had around a 5.7.  Then there's Henne at 4.9.  What's the difference?  A couple of tenths of a point.  Which if I'm translating this corrrectly is about 3/4's of a yard difference per pass attempt from Henne to an average QB.  And this means that Henne STINKS?  Players ANY/A will go up or down by several points from year to year depending on how successful his team is.  Does that mean that QB's go from being great QB's one year to lousy ones the next or vice versa?  If the stat really determines how good a QB is wouldn't you kinda expect that a QB's ANY/A would gradually get better every year, peak at somewhere in the middle of their career and then gradually tail off at the end again?  That's NOT what happens.  It fluctuates WILDLY from year to year in most cases. One year it's 8.2, the next it's 6.7 or vice versa.  What's that telling you?  It's telling you that it really doesn't tell you much about what the player will do the next year, it's only telling you whether or not he had a good year this year which can be dependant on a MILLION things, not just the talent level of the QB.

Hell it's not even ACCURATE for Henne this year because they have his stats WRONG.  He DIDN'T throw 19 INT's he only threw 18, but because the refs were brain dead that play OFFICIALLY he threw 19 INT's.  So because the refs made a mistake, Henne is a suckier QB?  Really?  Well then maybe some of those people above Henne just happened to get LUCKY and get better refs or perhaps Henne was LUCKY to get the refs he did and he's really even suckier than his ANY/A shows.

It's a stat.  That's all it is.  It doesn't prove anything except that Henne had a bad year.  No shit.  I don't really need the ANY/A to tell me that.  Does it mean that he'll be just as sucky next year?  Nope.  Not even a little.  It's NOT a predictor of future outcome, it's merely an indication of the type of year that the QB had this year.  We'll just have to wait till next year to find out if he's sucky again.
« Last Edit: January 19, 2011, 06:03:13 pm by Pappy13 » Logged

That which does not kill me...gives me XP.
fyo
Uber Member
*****
Posts: 7545


4866.5 miles from Dolphin Stadium


« Reply #51 on: January 19, 2011, 05:32:33 pm »

Do you require the same context when you say that Chris Johnson had a 2000-yard season

If you are going to use that 2000-number to argue that Chris Johnson is a better running back than random-other-back who had 1800 yards, yes.

Quote
What makes you think first downs aren't part of the metric?

They aren't. You quoted the formula. There were no "downs" in it.

Quote
That would seem to place an even bigger value on interceptions than the 45-yard number that you complained about.

Nice of you to pick and choose Wink. As I pointed out, it's only relevant in the comparison of what a touchdown is worth. Going by the numbers you supply, they are valuing a touchdown at 1.3 points and an interception at 2.9 points. It's the ratio that I'm arguing is significantly off.

Quote
I know enough about the statistical crunching to realize that we are both FAR out of our depth; while I don't pretend to know all the variables, I've read enough articles on the subject to recognize your evaluation as far too simplistic to be considered comparable.  Again, I suggest you read the article I linked in my previous post.

I've read just about every "football stat" site out there. My statistical background is fairly solid, certainly good enough to crunch the numbers these sites do.

Quote
if you want a formula that can be applied to every QB fairly easily, and still give more contextual value than simple passer rating, ANY/A is the way to go.

There is no more context in ANY/A than passer rating. Both have ZERO context.

ANY / A = (pass yards + 20*(pass TD) - 45*(interceptions thrown) - sack yards)/(passing attempts + sacks)

Passer Rating = 100 * [ (completions/attempt - 0.3)*5 + (yards/attempt - 3)*0.25 + (TD/attempt)*20 + 2.375-(INT/attempt)*25 ] / 6

In situations where none of the components of the passer rating are maxed out (there's a min and max for each), it's easy to see that yards/attempt count only a quarter in ANY/A and interceptions nearly twice (1.Cool times as much. Other than that, ANY/A counts all sacks AGAINST the quarterback, while ignoring completion percentage (and vice versa with passer rating, which ignores sacks, while including completion percentage).

Should interceptions really count for twice what they do in passer rating?
Should sacks really count against the quarterback? (Is this really something you want to fold into a single quarterback stat? Why not rushing then?)
Should completion percentage really be completely ignored?

Like I said, as a tool for quarterback evaluation, it seems like there's little to favor ANY/A over standard passer rating.
Logged
JVides
Uber Member
*****
Posts: 2915



« Reply #52 on: January 19, 2011, 10:29:07 pm »

^^^You guys need jobs...who has time for this shit?   Wink
Logged

"under wandering stars I've grown
by myself but not alone
I ask no one"
Metallica, "Wherever I may Roam"
tepop84
Uber Member
*****
Posts: 635


« Reply #53 on: January 20, 2011, 12:25:05 am »

take 2 qbs, QB A goes 30/31 for 300 yards with 1td/1int. He also takes 10 sacks for -100 yards.  QB B goes 30/41 for 300 yards with 1td/1int.  He doesn't take a sack.

It is clear that QB B is the better qb, however the qb rating for QB A is 104.3, where it has QB B is listed at 91.5. A's any/a is 4.375, and B is 6.7. 

Completion percentage is a really useless qb stat.
Logged
fyo
Uber Member
*****
Posts: 7545


4866.5 miles from Dolphin Stadium


« Reply #54 on: January 20, 2011, 06:05:32 am »

It is clear that QB B is the better qb, however the qb rating for QB A is 104.3, where it has QB B is listed at 91.5. A's any/a is 4.375, and B is 6.7. 

No, not clear at all. Seems like QB A had a horrible offensive line making his remarkable completion percentage all the more striking (and indicative of skill). In terms of helping his team win, those incompletions can really hurt the team -- how many punts would they lead to on average? (As opposed to rushing TDs, which aren't counted for the quarterback.)

QB B's stats could easily cover a losing effort against a "bend-don't-break" defense, that allowed him to go the first 50 yards and then incomplete passes forced a punt or settling for 3 points.

Yes, you could conjure up a scenario where QB B throws the ball away every time QB A was sacked... but, like I said before, it's easy to manufacture scenarios that make one stat or the other look bad -- especially since neither contain any CONTEXT.
« Last Edit: January 20, 2011, 06:07:17 am by fyo » Logged
badger6
Uber Member
*****
Posts: 1218



« Reply #55 on: January 20, 2011, 01:44:12 pm »

^^^You guys need jobs...who has time for this shit?   Wink

Bwahahahahahaaaaaaa
Logged
Doc-phin
Uber Member
*****
Posts: 1325


« Reply #56 on: January 20, 2011, 04:33:31 pm »

If this were a debate, I would be awarding most of the points to Tepop.

I have defended Thigpen against haters in the past and been luke warm as to promoting him.  But I have to say that from what I have seen he has developed a good bit even without getting proper practice reps.  I saw him throw a good bit of incompletions that hit receivers in the hands, numbers and helmets.  I only recall one interception that was his fault and that was against Pittsburg in 2009.  His O-line was in absolute disarray in the only start he got with us and it was on a short week.  As we all saw in that game, the defenders were on him as soon as he touched the ball.

Perhaps Tepop is simply a better QB evaluator than any of us.  He came across as nothing more than a Henne hater with an agenda but in reality his foresight was pretty accurate.  With the exception of Kolb or Hasselbeck, I can't say I would prefer someone else under center at this point.
Logged
masterfins
Uber Member
*****
Posts: 5557



« Reply #57 on: January 20, 2011, 05:29:25 pm »

I'd like to get that re-tread from the Colts.  As of yesterday the Colts hadn't started negotiations with P Manning yet, although Irsay says he will make him the highest paid player.  Ross outta throw a couple hundred million at Manning to try and tempt him, worse thing that could happen would be the Colts would have to shell out more money.  Oh...and maybe a little public embassment when he rejects the offer, but hey everyone's already forgotten about the last fiasco.
Logged
badger6
Uber Member
*****
Posts: 1218



« Reply #58 on: January 20, 2011, 06:53:36 pm »

If this were a debate, I would be awarding most of the points to Tepop.

I have defended Thigpen against haters in the past and been luke warm as to promoting him.  But I have to say that from what I have seen he has developed a good bit even without getting proper practice reps.  I saw him throw a good bit of incompletions that hit receivers in the hands, numbers and helmets.  I only recall one interception that was his fault and that was against Pittsburg in 2009.  His O-line was in absolute disarray in the only start he got with us and it was on a short week.  As we all saw in that game, the defenders were on him as soon as he touched the ball.

Perhaps Tepop is simply a better QB evaluator than any of us.  He came across as nothing more than a Henne hater with an agenda but in reality his foresight was pretty accurate.  With the exception of Kolb or Hasselbeck, I can't say I would prefer someone else under center at this point.

Even though the final outcome was different. Thigpen played better against the bears in a loss than Henne did in a win against the jets. If only they let him play the last few games of the year instead of just garbage time in the last game of the year that we were gonna lose anyhow !!!
Logged
Pappy13
Uber Member
*****
Posts: 8359



« Reply #59 on: January 21, 2011, 10:28:08 am »

Even though the final outcome was different. Thigpen played better against the bears in a loss than Henne did in a win against the jets.
Barely and don't forget that Henne's game was played in a downpour most of the game.

Thigpen: 17-29-187-0-1-6-(39) = 2.94 ANY/A and 63.4 Passer rating.
Henne: 5-18-69-1-0-5-(25) = 2.78 ANY/A and 53.3 Passer rating.

And for the year Henne played better.

Thigpen: 33-62-435-2-2-8-(50) = 4.79 ANY/A and 73.0 Passer rating.
Henne: 301-490-3301-15-18-30-(178) = 5.08 ANY/A and 75.4 Passer rating.

I don't see much reason to think that Thigpen is any more capable than Henne based on this year's stats.
« Last Edit: January 21, 2011, 10:32:50 am by Pappy13 » Logged

That which does not kill me...gives me XP.
Pages: 1 2 3 [4] 5 Print 
« previous next »
Jump to:  

The Dolphins Make Me Cry - Copyright© 2008 - Designed and Marketed by Dave Gray


Powered by SMF 1.1.21 | SMF © 2015, Simple Machines