Brunell vs. Bledsoe

[That Guy]

Originally Posted by Huddle

I'm very patient. Let's try this one more time.

There is one logical argument only: You cannot claim that your statistic is a measurement of a player's performance when that statistic is a combined measurement of the player's performance and other significant factors (You cannot measure A,B,C,D,E together and rely on it as a measurement of A).

The evidence of its unreliability is in the sharp rise and fall of of the stats of many players when they change teams or when a new coach uses them differently in their scheme.

That's it.

again, have you taken algebra?

X + Y + Z = 6
2X + Y + Z = 9
X = 3

the more data you have and the more equations you have the better able you are to judge the individual factors and figure out their roles in said equations.

[That Guy]

Quote:
when you have multiple variables, you make multiple equations and you CAN find out A, B, C or D individually from them. that's basic math.

Right, basic math with a false analogy. What we're talking about here is more like trying to solve an equation where there are no givens. We can think of five or six factors which can influence the statistic and we can't isolate and accurately measure any of them.

you can accurately measure all of it. the speeds, the times, the tackles, the fumbles, the knee bend, EVERYTHING. So your retort here is clearly wrong

Quote:
portis in washington still put up monsster numbers. its not like his stats got cut in half. situation does play a role, but like i said, the individual player plays a much bigger role on his own statline.

Probably but if you offered a percentage on how much to give the player's performance, you'd have to reach behind you to find it.

Well then, you've just admitted that stats aren't worthless right there. so why are you still going at it so hard?

[That Guy]

what parameters do you want on it? If you want you can treat each team as an equation and each year givess you 32 equations... you compile that data and you can data mine it any way you want.

using all the data to find the one you want, like yearly passing yards for a 10-6 season with 1500 yards rushing and a 10th ranked defene or whatever. Or use it to factor out the differences between a run heavy and pass heavy teams stats.

In the end the stats as they are work well enough that there's no really need to go into that though. you can look at sacks and the QBs escape ability (measured from bledsoe to mcnabb) and have some idea how much of that is on the OL and how much is on the QB.

you can look at fumbles/sack and fumbles/play (or per play hit, to avoid OL factoring in) for QBs to see their ball security averages (with alex smith being one of the worst).

what exactly are you looking to prove and how in depth does it have to be? stats are a measure of production, and when people talk about who's better, they generally mean in terms of production. jim brown is good cause he was productive and he was able to generate wins. Manning is productive because he OD's on film study and produces crazy passing stats.

if manning had moss and cooley and the junk WRs we had last year, he'd put up better numbers than brunell. his stats prove that by the crazy difference in TD/INT, TDs, yards, completion % etc.

If two players are close you can argue the stats may be misleading, but in many cases they point out the obvious quite well. manning isn't the lottery winner of a good scheme, he's the driver. if he had junk WRs his stats wouldn't be as good, but they'd clearly be better than putting brunell or vick or alex smith in the same situation.

[Huddle]

That Guy

what parameters do you want on it? If you want you can treat each team as an equation and each year givess you 32 equations... you compile that data and you can data mine it any way you want.

I'm sure you can manipulate the numbers and come up all sorts of statistics but I hope you are going to show me how you could produce some that are useful and reliable.

In the end the stats as they are work well enough that there's no really need to go into that though. you can look at sacks and the QBs escape ability (measured from bledsoe to mcnabb) and have some idea how much of that is on the OL and how much is on the QB.

Explain how you do that, please. I'm skeptical.

if manning had moss and cooley and the junk WRs we had last year, he'd put up better numbers than brunell. his stats prove that by the crazy difference in TD/INT, TDs, yards, completion % etc.

In order to prove this, you'd have to assign a weight as a percentage to both the Redkins "support package" and the Colts support package. And, if you were able to do this accurately, you could get a very accurate grade of the quarterback's performance isolated from other factors.

I don't believe you can do it.

How do you grade interceptions, for example. Suppose a QB has 17 for the year. It seems to me that to grade fairly, we need to know when and how they happened. If his team was poor defensively, and their opponents ran out to big leads, then we'd need to make an adjustment to that number for the Hail Marys and other INTs thrown in desperate situations.

If his coach liked to go deep often, or his receivers didn't fight for passes, or if his offensive line didn't protect him...all of these things and others would factor into his total.

So, if A represents INTs that are primarily the fault of the QB, then we have:

A + B + C + D + E = 17 What's the value of A?

[offiss]

Originally Posted by Huddle

That Guy



I'm sure you can manipulate the numbers and come up all sorts of statistics but I hope you are going to show me how you could produce some that are useful and reliable.



Explain how you do that, please. I'm skeptical.



In order to prove this, you'd have to assign some arbitrary weight as percentage to both the Redkins "support package" and the Colts support package. And, if you were able to do this, you could get a very accurate grade of the quarterback's performance isolated from other factors.

I don't believe you can do it.

How do you grade interceptions, for example. Suppose a QB has 17 for the year. It seems to me that to grade fairly, we need to know when and how they happened. If his team was poor defensively, and their opponents ran out to big leads, then we'd need to make an adjustment to that number for the Hail Marys and other INTs thrown in desperate situations.

If his coach liked to go deep often, or his receivers didn't fight for passes, or if his offensive line didn'tprotect him...all of these things and others would factor into his total.

So, if A represents INTs that are primarily the fault of the QB, then the equation is:

A + B + C + D + E = 17 What's the value of A?

Nice job Huddle, don't get to worked up in the argument though, guy's who put all their stock in stats usually can't break the game down, they know what they believe, don't confuse them with reality.

[STPainmaker]

What the hell are you guys fighting about. Maybe you should have a meeting and discuss the finer points of statistical analysis. I hope you do this in a sound proof room and manage to cripple each others hands so we won't be privy to any of the horrendously stupid, and ridiculous details.

[Huddle]

Originally Posted by STPainmaker

What the hell are you guys fighting about. Maybe you should have a meeting and discuss the finer points of statistical analysis. I hope you do this in a sound proof room and manage to cripple each others hands so we won't be privy to any of the horrendously stupid, and ridiculous details.

Are you angry because you hate details or because the discussion went over your head? In any case, don't blame us, blame the person who forced you to sit there and read it.

[That Guy]

Originally Posted by offiss

don't confuse them with reality.

btw, how's portis's roid binge going?

[offiss]

Originally Posted by That Guy

btw, how's portis's roid binge going?

No different than Bonds, Sheffield, and Giambi, as well as the rest of the NFL.

[Bozzy]

Brunell's completion % and sacks allowed are lower than Bledsoe's numbers because Brunell is obviously better at throwing the ball away and avoiding the sack, thus lowering the completion %.

[dmek25]

hey offiss,how the hell are you?did you wish your boy ramsey good luck?

[offiss]

Originally Posted by dmek25

hey offiss,how the hell are you?did you wish your boy ramsey good luck?

Well he's moving to my neighborhood now, unfortunatly I really have no idea on how this new Jet's coaching staff is going to pan out, I was hoping he would winde up in a better situation than the Jets.

[That Guy]

Not being able to understand how it is done does not mean it can't be done.

-Where did I say that it did?

how about right here:

-In order to prove this, you'd have to assign some arbitrary weight as percentage to both the Redkins "support package" and the Colts support package. And, if you were able to do this, you could get a very accurate grade of the quarterback's performance isolated from other factors.

I don't believe you can do it.

stats and algebra say you can (and there's MORE than enough data to set it up, though it could take weeks or months to make it accurate, its already been done in both baseball and the nba). stats are used to show production and give you an idea of how good players are that you don't have time to watch...

Good players tend to end up with good stats, so in general, the system does work. Sometimes good players have bad stats, but bad players rarely have good numbers. If you can show me an example of a terrible player racking up crazy stats, please do. Otherwise the whole basis for stats still holds.

[12thMan]

Wow, I just got back on...this thread has really taken on a life of it's own.

Anyway, I was hoping someone could confirm the rumor the Spurrier has fled for the CFL?

[Huddle]

That Guy and others....

Since you have avoided a direct confrontation with the argument I made and the evidence I offered, I'll assume at this point that you can't find a counter argument.

I realize that it's easier for you to put words in my mouth and then trash the things I never said, but for the record:

I did not state that all statistics are useless.

I did not state that the statistics used by pro teams are useless.

I did not state that baseball statistics are useless
.
I did not state that, even given additional new data, it would still be impossible to grade individual performance.

I said that the common football statistics of the kind so often used in this forum and others are almost useless.

I gave one logical argument for my position and I supplied evidence that the stats are unreliable.

As stated in one of my earlier posts:

There is one logical argument only: You cannot claim that your statistic is a measurement of a player's performance when that statistic is a combined measurement of the player's performance and other significant factors (You cannot measure A,B,C,D,E together and rely on it as a measurement of A).

The evidence of its unreliability is in the sharp rise and fall of of the stats of many players when they change teams or when a new coach uses them differently in their scheme. If the stats were a reliable measure of individual performances, this evidence would not exist.

Unless, there are new arguments on point, I'll rest my case.