Marvel Mystery Run on Heritage

[sfilosa]

Don't take offense but PLEASE:

 

There is no way you would take a CGC 9.2 book that you felt was a 9.4, and crack it out (and not resubmit) and sell it raw as a 9.4. That makes ZERO SENSE.

 

 

--------------------------------------------------------------------------------

 

 

 

No offense taken.

 

But I stand by my position as to what I would or would not do under the circumstances we have been discussing.

 

And the reasoning is simple - integrity and principle. I prefer not to allow financial motivation, and dare I say greed, compromise either. And I think CGCresubmits in the scenario discussed above does just that, as well as undercut the credibility of CGC. I simply don't want to be a party to that.

 

I understand your position.

 

Then wouldn't it be better to leave the book in the slab, call for the graders notes (to make sure there are no hidden defects) and say the book looks like a 9.4. By leaving it in the holder, you give confidence to a prospective buyer that the book is at least a CGC 9.2 and has no restoration.

 

If you deslab it, and sell it as a 9.4 (without mentioning that CGC called it a 9.2), is that really any better then resubmitting the book? If the new buyer submits the book and CGC still gives it a 9.2, then your considered an overgrader, and since the difference in price between a 9.2 and 9.4 is huge, I would think you would have an unhappy buyer.

[FFB]

What I would do, if I disagreed with a CGC grade, is crack the book out and advertise it as the grade I think it is. Now of course I realize that perhaps my saying a book is a 9.4 does not have the same "weight" as a CGC 9.4 and so it might not garner the same price.

 

Don't take offense but PLEASE:

 

There is no way you would take a CGC 9.2 book that you felt was a 9.4, and crack it out (and not resubmit) and sell it raw as a 9.4. That makes ZERO SENSE.

 

No offense taken.

 

But I stand by my position as to what I would or would not do under the circumstances we have been discussing.

 

And the reasoning is simple - integrity and principle. I prefer not to allow financial motivation, and dare I say greed, compromise either. And I think CGCresubmits in the scenario discussed above does just that, as well as undercut the credibility of CGC. I simply don't want to be a party to that.

 

If CGC screws me by undergrading a book, I have no problem resubmitting it for a shot at a better (and more accurate) grade. Why should I pay the price because they were harsher than usual on the day they graded the book?

 

And undergrading doesn't mean CGC lacks credibility. Only the least sophisticated buyer imaginable would believe that CGC is 100% consistent on every single day, or hold it against CGC if they grade a book 9.6 one day and 9.8 the next. There isn't a grader in the world who can claim that level of consistency. We all overgrade or undergrade sometimes. The owner of the book shouldn't be handcuffed just because he got a slightly undergraded book.

[Zanarkand]

Very cool analysis. I question two of your assumptions though. One that doesn't make much of a difference but also one that does.

 

You shouldn't leave out all the minor cases with low probabilities of occurring (such as all three graders being way off) because they add up - in CGC's favor.

 

My assumption is that:

 

X-- = -2 (i.e. 2 grade steps low)

X-1 = -1

X = 0

X+1 = 1

X++ = 2

 

Also I assume that the "final" grade is a strict numerical average of the above, rounded to the nearest integer.

 

So X--, X, X+ for example would work out to X. Meaning that one grader way undergraded, one slightly overgraded, and one was correct, but the final grade works out to be "correct" anyway. Just by accident, but it still counts towards CGCs "correct" percentage.

 

Also, I think if we're calling X the absolute "correct" grade, then everything else should be termed an absolute as well. So X+1 and X-1 aren't "maybe over/undergraded", they are "definitely over/undergraded 1 step". Or if you prefer, "final grade assigned is definitely over/under the 'correct' grade by one step", as compared to your absolute X.

 

Anyway, using your probability of an individual grade being 75% accurate and no bias, I got:

 

P(2 step undergrade) = 0.05%

P(1 step undergrade) = 10.65%

P(correct) = 78.60%

P(1 step overgrade) = 10.65%

P(2 step overgrade) = 0.05%

 

Fiddling with the results a little:

 

Probability the CGC is correct on purpose = P(G1 is correct) * P(G2 is correct) * P(G3 is correct) = 75% * 75% * 75% = 42.2%

 

Probability that CGC is correct just because the average of all three graders happened to work out to X, even though some or all of them were wrong = P(CGC is correct) - P(CGC is correct on purpose) = 78.6% - 42.2% = 36.4%

 

Probability the CGC is wrong = P(not correct) = 21.4%

 

Probability that CGC is wrong but who cares = P(2 step undergrade) + P( 1 step undergrade) = 10.3%

 

Probability of a desirable outcome = P(correct) + P(2 step undergrade) + P(1 step undergrade) = 89.30%

 

Even though the probability of a desirable outcome is good, I find it troubling that by this analysis CGC is correct "on purpose" only a little more often than they are correct by accident (42.4% vs 36.4%), and also that they slap a "wrong" grade on a book (high or low) 21.4% of the time, or about 1 in 5 slabs.

[sfilosa]

While I understand the analysis, let me throw a wrench into everything:

 

The final grader can override the other two graders.

 

Basically, if the Pre-Grader says a book is 8.5 and the other grader says 8.5 but the finalizer says 9.0, then the book can be a 9.0. All three graders don't necessarily have the same weight.

 

 

THAT SAID:

FFB and I are in agreement. There is nothing wrong with resubmitting a book. Ignore the pressing issue, NO ONE should have a problem "APPEALING" a GRADING RULING.

 

I thought the lawyers would like that analogy.

[Scrooge]

Zanarkand, let me preface this by thanking you for taking the time to look through it. So far, you are the only one. Let me try to address your comments below:

 

One that doesn't make much of a difference.

 

You shouldn't leave out all the minor cases with low probabilities of occurring (such as all three graders being way off) because they add up - in CGC's favor.

 

My assumption is that:

 

X-- = -2 (i.e. 2 grade steps low)

X-1 = -1

X = 0

X+1 = 1

X++ = 2

 

Actually from my original posting, I said that: "Then they can deviate by more than one grade but this is unlikely and represented by a lower probability 4% off up by more than 1 grade and 4% off down by more than 1 grade." meaning that the X-- and X++ do not mean exactly by two steps but at least by two steps so that I do span all the universe with my probabilities. As you said, it wouldn't have mattered much anyway.

 

Also I assume that the "final" grade is a strict numerical average of the above, rounded to the nearest integer.

 

So X--, X, X+ for example would work out to X.

 

This is as I have it in my grid, yes (mostly, see below for additional comment on that particular topic)

 

Meaning that one grader way undergraded, one slightly overgraded, and one was correct, but the final grade works out to be "correct" anyway. Just by accident, but it still counts towards CGCs "correct" percentage.

 

Also, I think if we're calling X the absolute "correct" grade, then everything else should be termed an absolute as well. So X+1 and X-1 aren't "maybe over/undergraded", they are "definitely over/undergraded 1 step". Or if you prefer, "final grade assigned is definitely over/under the 'correct' grade by one step", as compared to your absolute X.

 

Maybe you mis-read what I meant to say. Look at the grid again but when the average grades end up being NOT X, the book is ALWAYS listed as either Overgraded or Undergraded dependent on the scenario.

 

What I think is happening is that you are mis-reading my Maybe Overgraded and Maybe Undergraded. These qualifications do not arise when the arithmetic average of the 3 graders is different from X BUT they occur when at least 2 graders agree and assign the correct grade of X and YET the third grader assign a higher or lower grade. Overall, CGC is actually Correct in its grades but from A COLLECTOR's POINT of VIEW, the book might be undergraded if the dissenting third grader gave a X+ or X++ even though the grade is correct. But we agree, my MU and MO are in your modified model part of the Correct category.

 

Anyway, using your probability of an individual grade being 75% accurate and no bias, I got:

 

P(2 step undergrade) = 0.05%

P(1 step undergrade) = 10.65%

P(correct) = 78.60%

P(1 step overgrade) = 10.65%

P(2 step overgrade) = 0.05%

 

Fiddling with the results a little:

 

Probability the CGC is correct on purpose = P(G1 is correct) * P(G2 is correct) * P(G3 is correct) = 75% * 75% * 75% = 42.2%

 

Probability that CGC is correct just because the average of all three graders happened to work out to X, even though some or all of them were wrong = P(CGC is correct) - P(CGC is correct on purpose) = 78.6% - 42.2% = 36.4%

 

Probability the CGC is wrong = P(not correct) = 21.4%

 

[...]

 

Even though the probability of a desirable outcome is good, I find it troubling that by this analysis CGC is correct "on purpose" only a little more often than they are correct by accident (42.4% vs 36.4%)

 

Actually I find that result more conforting than disturbing because it shows in clear terms the advantage of having the 3 graders' system. What is disheartening (assuming our numbers are correct) is the improvement ratio from the system. By this I mean we start with graders with 75% accuracy and the system only marginally raise that probability to 78.6%. The improvement ratio is even smaller for higher prior beliefs, i.e., if we assume that individual graders are better on average.

 

For that matter, I don't necessarily agree with your numbers, I would estimate that Prob(CGC is Correct) is 88% and Prob(CGC is Wrong) = 12% and the improvement from the system is much greater in my estimation. I believe that we differ because you stick closely to the rounding to the nearest integer rule while I am more generous in favor of CGC. I feel that in the case below:

 

G1 = X, G2 = X and G3 = X++ you give the CGC given grade as X+1 hence adding to the probability of CGC being Wrong while I give it a CGC given grade of X adding to the weight of CGC being right. I made that judgement call because there is an extra step for the final grade approval and having 2 graders at X would increase the likelihood the grade assigned would be the one on which 2 graders agreed. Debatable decision but that was my working assumption. ALSO, I feel justified in this because this goes back to Steve's comment (with which I agree) about grades not necessarily having the same weight and I was accounting for this by assigning final grades in a heuristic manner instead of abiding by a fast arithmetic average rule.

 

Now, as Mark pointed out earlier, unless we run a fairly large size sampling of grades we won't know for sure how many books in % are mis-graded but, Mark, at least having build this model, we can recover the priors (= probabilities of accuracy of the average CGC grader) which to me is a more important number and a number that a random sampling would not allow us to infer.

[Zanarkand]

While I understand the analysis, let me throw a wrench into everything:

 

The final grader can override the other two graders.

 

Basically, if the Pre-Grader says a book is 8.5 and the other grader says 8.5 but the finalizer says 9.0, then the book can be a 9.0. All three graders don't necessarily have the same weight.

 

A good point. Scrooge & I are playing with a basic "all things being equal" sort of scenario. In addition to that, there really is a bias that we haven't accounted for - one would hope they would tend to under, not over, grade.

[adamstrange]

What I would do, if I disagreed with a CGC grade, is crack the book out and advertise it as the grade I think it is. Now of course I realize that perhaps my saying a book is a 9.4 does not have the same "weight" as a CGC 9.4 and so it might not garner the same price.

 

Don't take offense but PLEASE:

 

There is no way you would take a CGC 9.2 book that you felt was a 9.4, and crack it out (and not resubmit) and sell it raw as a 9.4. That makes ZERO SENSE.

 

No offense taken.

 

But I stand by my position as to what I would or would not do under the circumstances we have been discussing.

 

And the reasoning is simple - integrity and principle. I prefer not to allow financial motivation, and dare I say greed, compromise either. And I think CGCresubmits in the scenario discussed above does just that, as well as undercut the credibility of CGC. I simply don't want to be a party to that.

 

Isn't the credibility issue with CGC and not with the re-submitter? If CGC always returned the same grade, would anyone re-submit?

[tth2]

Zanarkand, let me preface this by thanking you for taking the time to look through it. So far, you are the only one. Let me try to address your comments below:

 

One that doesn't make much of a difference.

 

You shouldn't leave out all the minor cases with low probabilities of occurring (such as all three graders being way off) because they add up - in CGC's favor.

 

My assumption is that:

 

X-- = -2 (i.e. 2 grade steps low)

X-1 = -1

X = 0

X+1 = 1

X++ = 2

 

Actually from my original posting, I said that: "Then they can deviate by more than one grade but this is unlikely and represented by a lower probability 4% off up by more than 1 grade and 4% off down by more than 1 grade." meaning that the X-- and X++ do not mean exactly by two steps but at least by two steps so that I do span all the universe with my probabilities. As you said, it wouldn't have mattered much anyway.

 

Also I assume that the "final" grade is a strict numerical average of the above, rounded to the nearest integer.

 

So X--, X, X+ for example would work out to X.

 

This is as I have it in my grid, yes (mostly, see below for additional comment on that particular topic)

 

Meaning that one grader way undergraded, one slightly overgraded, and one was correct, but the final grade works out to be "correct" anyway. Just by accident, but it still counts towards CGCs "correct" percentage.

 

Also, I think if we're calling X the absolute "correct" grade, then everything else should be termed an absolute as well. So X+1 and X-1 aren't "maybe over/undergraded", they are "definitely over/undergraded 1 step". Or if you prefer, "final grade assigned is definitely over/under the 'correct' grade by one step", as compared to your absolute X.

 

Maybe you mis-read what I meant to say. Look at the grid again but when the average grades end up being NOT X, the book is ALWAYS listed as either Overgraded or Undergraded dependent on the scenario.

 

What I think is happening is that you are mis-reading my Maybe Overgraded and Maybe Undergraded. These qualifications do not arise when the arithmetic average of the 3 graders is different from X BUT they occur when at least 2 graders agree and assign the correct grade of X and YET the third grader assign a higher or lower grade. Overall, CGC is actually Correct in its grades but from A COLLECTOR's POINT of VIEW, the book might be undergraded if the dissenting third grader gave a X+ or X++ even though the grade is correct. But we agree, my MU and MO are in your modified model part of the Correct category.

 

Anyway, using your probability of an individual grade being 75% accurate and no bias, I got:

 

P(2 step undergrade) = 0.05%

P(1 step undergrade) = 10.65%

P(correct) = 78.60%

P(1 step overgrade) = 10.65%

P(2 step overgrade) = 0.05%

 

Fiddling with the results a little:

 

Probability the CGC is correct on purpose = P(G1 is correct) * P(G2 is correct) * P(G3 is correct) = 75% * 75% * 75% = 42.2%

 

Probability that CGC is correct just because the average of all three graders happened to work out to X, even though some or all of them were wrong = P(CGC is correct) - P(CGC is correct on purpose) = 78.6% - 42.2% = 36.4%

 

Probability the CGC is wrong = P(not correct) = 21.4%

 

[...]

 

Even though the probability of a desirable outcome is good, I find it troubling that by this analysis CGC is correct "on purpose" only a little more often than they are correct by accident (42.4% vs 36.4%)

 

Actually I find that result more conforting than disturbing because it shows in clear terms the advantage of having the 3 graders' system. What is disheartening (assuming our numbers are correct) is the improvement ratio from the system. By this I mean we start with graders with 75% accuracy and the system only marginally raise that probability to 78.6%. The improvement ratio is even smaller for higher prior beliefs, i.e., if we assume that individual graders are better on average.

 

For that matter, I don't necessarily agree with your numbers, I would estimate that Prob(CGC is Correct) is 88% and Prob(CGC is Wrong) = 12% and the improvement from the system is much greater in my estimation. I believe that we differ because you stick closely to the rounding to the nearest integer rule while I am more generous in favor of CGC. I feel that in the case below:

 

G1 = X, G2 = X and G3 = X++ you give the CGC given grade as X+1 hence adding to the probability of CGC being Wrong while I give it a CGC given grade of X adding to the weight of CGC being right. I made that judgement call because there is an extra step for the final grade approval and having 2 graders at X would increase the likelihood the grade assigned would be the one on which 2 graders agreed. Debatable decision but that was my working assumption. ALSO, I feel justified in this because this goes back to Steve's comment (with which I agree) about grades not necessarily having the same weight and I was accounting for this by assigning final grades in a heuristic manner instead of abiding by a fast arithmetic average rule.

 

Now, as Mark pointed out earlier, unless we run a fairly large size sampling of grades we won't know for sure how many books in % are mis-graded but, Mark, at least having build this model, we can recover the priors (= probabilities of accuracy of the average CGC grader) which to me is a more important number and a number that a random sampling would not allow us to infer.

Hulk's head hurts.

[Silver Surfer]

I tried to read the very detailed and well presented numerical analysis but I fell asleep, just like I use to in algebra 12.

[adamstrange]

 

My thoughts too.

 

Sorry, Scrooge, but at then end of a really long day at work, this is definitely not what I hoped to find on a comics forum.

[Scrooge]

 

My thoughts too.

 

Sorry, Scrooge, but at then end of a really long day at work, this is definitely not what I hoped to find on a comics forum.

 

Professional bias creeping in. I deal with these analyses every day so it's all routine to me. I'll blame it on Zanarkand for reviving this apparently all around painful post.

[tth2]

 

My thoughts too.

 

Sorry, Scrooge, but at then end of a really long day at work, this is definitely not what I hoped to find on a comics forum.

 

Professional bias creeping in. I deal with these analyses every day so it's all routine to me. I'll blame it on Zanarkand for reviving this apparently all around painful post.

Scrooge, I wasn't making fun of you or your analysis. More making fun of myself in completely blanking out as soon as I saw rows and rows of statistical analysis.

[Zanarkand]

 

My thoughts too.

 

Sorry, Scrooge, but at then end of a really long day at work, this is definitely not what I hoped to find on a comics forum.

 

Professional bias creeping in. I deal with these analyses every day so it's all routine to me. I'll blame it on Zanarkand for reviving this apparently all around painful post.

 

 

 

 

(fun though, huh?)

[esquirecomics]

Now, as Mark pointed out earlier, unless we run a fairly large size sampling of grades we won't know for sure how many books in % are mis-graded but, Mark, at least having build this model, we can recover the priors (= probabilities of accuracy of the average CGC grader) which to me is a more important number and a number that a random sampling would not allow us to infer.

 

Guys, I will personally thank you for all the hard work you did. I honestly don't have a clue what it means and probably never will (which is why I went to law school), but I will cite to it as support for every and any argument I make on these boards from now on!!!!

 

Pressing is bad, and here is why = G1 = X, G2 = X and G3 = X++

 

Resubmits are bad, and here is why = G1 = X, G2 = X and G3 = X++

 

That playboy model should have really gone out with me in college, and here is why = G1 = X, G2 = X and G3 = X++

 

This works beautifully!!

 

Seriously, thanks for the effort.

[sfilosa]

Resubmits aren't bad.

 

As I said before, think of it as APPEALING a GRADE RULING.

 

Regarding Pressing, I'm fine with your opinion (but disagree).

 

 

Are you bidding on anything in Heritage's Signature Auction (you don't actually have to mention any particular books).

[esquirecomics]

Resubmits aren't bad.

 

As I said before, think of it as APPEALING a GRADE RULING.

 

Actually, might be more appropriate to call it a MOTION FOR RECONSIDERATION.

 

Are you bidding on anything in Heritage's Signature Auction (you don't actually have to mention any particular books).

 

Thinking about it. Some fantastic books to be sure. I'm salivating. However, my wife and I just bought an investment condo in DC for $$$$$ so I may need to take a break from buying (unless, of course, you forumites buy some books from me!!).

[sfilosa]

Thinking about it. Some fantastic books to be sure. I'm salivating. However, my wife and I just bought an investment condo in DC for $$$$$ so I may need to take a break from buying (unless, of course, you forumites buy some books from me!!).

 

The market is very hot in DC now, so I'm sure it cost you a pretty penny.

 

The thing with Heritage's Signature Auctions is there are usually a few key / special books in every auction, for almost everyones taste (i.e. GA, SA, BA, OA Collector). Getting them for the right price is a whole different issue, so if you don't bid / win any this time, then there will be something else to grab your attention next time.

 

I would say that for a Timely Collector, this would probably be one of their best auctions in years, but for other collectors GA DC's, EC's, etc. they always have good stuff.

 

Good Luck.

[tth2]

That playboy model should have really gone out with me in college, and here is why = G1 = X, G2 = X and G3 = X++

Bad news, Mark, I ran this particular problem through the computer and it said this was the appropriate equation:

 

M(AR)/K + Z*AI+D = G+E/E*K

[Silver Surfer]

That playboy model should have really gone out with me in college, and here is why = G1 = X, G2 = X and G3 = X++

Bad news, Mark, I ran this particular problem through the computer and it said this was the appropriate equation:

 

M(AR)/K + Z*AI+D = G+E/E*K

 

[shrunkenhead]

Mark, I feel your pain, my wife and I just moved into a condo in Miami Beach in May, and the rates are brutal, man, brutal!!

 

I have had to shrink the comic book budget significantly, too. Suddenly VG+ looks nicer than ever...and ooooohh look at those pretty moderns.