Question for Valiantman (& anyone statistically minded)

[JohnBull]

Firstly, let me congratulate you on your site and thank you for it too. It's an amazing accomplishment and I am astounded by the curmudgeons on these boards who have stolen your 4th star. Given the work you've put into this and also into answering all the queries put your way, if anyone deserves a high rating, you do. Kudos also to CGC for giving you access to the information.

 

My particular interest in using your work is for analysing percentile breakdowns of particular grades of specific issues (within a run). What I'd like to be able to do is enter a run of issues (say FF 131-135) and a run of grades (maybe 9.4 to 9.8) and then see what percentage of each issue falls within each grade and also the total number submitted for each issue.

 

I realise that I can actually calculate this with the current set up, but the existing easy visual charts only list the percentages for each grade when looking at the run as a whole.

 

I tend to think that these percentages are where the meat and drink of the census analysis is to be found. There may well be several hundred Hulk 180's in 9.6 still to be graded, but I doubt that the percentage in that grade will vary very much from the current one of 3.62%. Ultimately, it's these percentages that will indicate the scarcity of any particular book in high grade.

 

If there were any way you could make some software tweaks, I'd be very appreciative. If the facility already exists but I'm too dense to see it...

 

Now some general questions for anyone statistically minded.

 

i) How many submissions do you think is the minimum needed in order to get a meaningful result of such analysis ? When I say meaningful result, I'm looking for something that is in the order of 95% accurate (ie: for Hulk 181 in 9.6, I think the 3.62% figure will be accurate to within plus or minus 5% - 3.44% to 3.80%). My gut feeling is that about 50 submissions are needed.

 

ii) Do you think that this analysis is only valid for comics of a certain age ? Is there any reason to think say, that the submission pattern for books from the 1940's is likely to be different to that of the 1950's ? I have a suspicion that this work might be best suited to books from the 1970's up, but I have nothing concrete to base this on. Ultimately, it might be interesting to see a breakdown of all books by a specific publisher in certain timeframes (ie: what is the number of 9.8's that exist for all Marvels from 1968-1975, in order to compare this with any specific issue's figures).

 

iii) Another gut feeling I have is that it should be possible to get meaningful results from a small run of issues, where the total number of submissions is sufficiently great even if the number for any particular book isn't. If I want to look at the Incredible Spiderthing issues 51-55, there may only be 10 submissions for each, but that total of 50 should provide a sufficient base to look at that run and still provide information for each specific issue. Does this seem reasonable ? If so, what do you think would be the minimum average number of submissions per issue needed ? If there were 10 each of FF 131-135 that might be ok, but if there were 5 each of FF 141-150, would that still be enough ? I tend to think yes to the first but no to the second, but I'm still only going by feel.

[valiantman]

Kudos also to CGC for giving you access to the information.

Just in case anyone thinks otherwise...

I use the same methods to look up each comic in the census as everyone does.

It's just that I wrote a program to look up every comic and store the results.

(There are about 39,000 comics in the census... so I do 39,000 regular look-ups.)

 

What I'd like to be able to do is enter a run of issues (say FF 131-135) and a run of grades (maybe 9.4 to 9.8) and then see what percentage of each issue falls within each grade and also the total number submitted for each issue.

Well... the system works for runs... say FF 131-135...

and it also works for single issues... just FF 131, for example.

The drop-down choices on the pages are for the dates,

but you can see in the URL (web address) where there are two variables

named "issuestart" and "issueend". Change "issuestart" to 131 and "issueend" to 131,

and you get a "full analysis" of just issue 131.

 

Just let me know if there is something else you'd like to see... or if that's what you needed.

Most enhancements don't take very long to code... so I can give them a shot!

 

i) How many submissions do you think is the minimum needed in order to get a meaningful result of such analysis ? ...I'm looking for something that is in the order of 95% accurate (...accurate to within plus or minus 5%). My gut feeling is that about 50 submissions are needed.

I'm not a statistician, but I am dangerous with stats from time to time...

According to this "Sample Size Calculator" that I found online...

You need about 375 total examples if you're looking for numbers

that are 95% confidence with a +/- 5% interval for the results.

(Set confidence level to 95%, confidence interval to '5', and population

to the number of books estimated to be in existence (or the print run)...

anywhere from 1,000 to 10,000,000... the result is always between 275 and 385.)

Sample Size Calculator

Note: This system assumes a RANDOM sample is being selected.

Since people don't generally submit their books at random,

because they will usually submit their best copy as opposed to "any" copy...

the universe is NOT truly random... but it is fairly consistent because

everyone generally follows that same rule.

Additionally, it should be noted that you requested 95% confidence in +/- 5% intervals.

If you said 98%, 99% or +/- 2%, or +/- 1%... the sample size would definitely increase.

 

ii) Do you think that this analysis is only valid for comics of a certain age ? Is there any reason to think say, that the submission pattern for books from the 1940's is likely to be different to that of the 1950's ?

The most important point in comparing comics is "motivation".

If there is no motivation to submit middle-grade or low-grade copies

of a particular issue... then they are not represented in the data.

High-dollar books tend to be submitted more in middle- and low-grade

than moderate dollar books. Low dollar books are almost never low-grade submissions.

 

The particular impact would be seen most easily on a book like Hulk 181...

since it has been submitted at all levels of the CGC scale...

but that is not true for "nearby" Hulk 179 or Hulk 183... because of the motivation.

There's nothing drawing mid- and low-grade copies of 179 and 183 into the sample.

When you examine the percentages... it would of course be misleading to say that

Hulk 181 is only high-grade (9.6 or better) 4.3% of the time,

while Hulk 179 is high-grade (9.6 or better) 35.7% of the time... based on CGC.

Obviously, only the highest-grade copies of Hulk 179 are being submitted,

since we don't have any logical, non-CGC-based reason to believe

Hulk 179's are 8 times "easier" in high-grade than Hulk 181, on average.

Hulk 179 is from the same time period, same company, and arguably,

they've probably been "less protected/sheltered" than Hulk 181 through the years.

It's a tough call to try to use percentages for "time-frames"

when key and non-key books are in those age groups.

 

iii) Another gut feeling I have is that it should be possible to get meaningful results from a small run of issues, where the total number of submissions is sufficiently great even if the number for any particular book isn't. If I want to look at the Incredible Spiderthing issues 51-55, there may only be 10 submissions for each, but that total of 50 should provide a sufficient base to look at that run and still provide information for each specific issue. Does this seem reasonable ?

When none of the run is a "key issue", or has any distinguishing feature...

your suggestion should be a valid method. However, it would probably require a total

number of submissions in the 250 to 350 range to be statistically strong.

So, if you want to check 5 issues, you'll need 50 to 70 of each...

if you want to check 20 issues, you might only need 12 to 18 of each.

If you go beyond 20 issues, you need to ask if the time period or "run" is

still fundamentally the "same" for all of the books being aggregated.

 

Whew, if you made it this far... I salute you.

[joe_collector]

Do you think that this analysis is only valid for comics of a certain age ?

 

One area to watch out for when looking at Census stats is the 1970-74 range, especially the non-Keys. These fall outside the lower cost Modern tier, yet are not usually valuable enough in CGC 9.2-9.4 to warrant the risk.

 

For example, I could send in 3 NM copies of X-Men 108 for the same price as 1 Captain America 180, Fantastic Four 121, or Captain Marvel 33, along with tons of other "personal faves" I'd love to own CGC'd. I have quite a few NM 20-cent Marvels (my favorite era) that would be down at CGC, if not for the ultra-high grading costs.

 

It doesn't take a genius to realize which books will be more highly-submitted, and which ones will have artificially lower Census numbers compared to their raw NM supply.

[mrwoogieman]

20-cent Marvels kick !

[CentaurMan]

The only way to get a significant sample size for much older comics, say 1940 and earlier, is to define a range, say Action Comics #2-#20, or all Centaur comics published in 1938. Unfortunately, this is where the site breaks down, as more than 2/3rds of the CGC slabs are graded 6.0 or less and are aggregated on the site into one data point.

 

Of course, that's when it also becomes clear that the site was designed expressly for the ultra high grade collector. There's nothing wrong with that. Just pointing out that even when you do manage to get a statistically significant sample size of early Golden Age books, the info provided from the site isn't granular enough to be of much use.

 

Still, very cool site. Majors kudos for setting it up.

[JohnBull]

One area to watch out for when looking at Census stats is the 1970-74 range, especially the non-Keys. These fall outside the lower cost Modern tier, yet are not usually valuable enough in CGC 9.2-9.4 to warrant the risk.

 

For example, I could send in 3 NM copies of X-Men 108 for the same price as 1 Captain America 180, Fantastic Four 121, or Captain Marvel 33, along with tons of other "personal faves" I'd love to own CGC'd. I have quite a few NM 20-cent Marvels (my favorite era) that would be down at CGC, if not for the ultra-high grading costs.

 

It doesn't take a genius to realize which books will be more highly-submitted, and which ones will have artificially lower Census numbers compared to their raw NM supply.

 

True, but the idea is to catch the ratio of the high grades. If 75% of books from this era catch a 9.2+ (since people may not sub unless they really fancy their chances), the grade ratios on specific issues compared to everything else published in that particular timeframe, could still be relevant.

[JohnBull]

The only way to get a significant sample size for much older comics, say 1940 and earlier, is to define a range, say Action Comics #2-#20, or all Centaur comics published in 1938. Unfortunately, this is where the site breaks down, as more than 2/3rds of the CGC slabs are graded 6.0 or less and are aggregated on the site into one data point.

 

Of course, that's when it also becomes clear that the site was designed expressly for the ultra high grade collector. There's nothing wrong with that. Just pointing out that even when you do manage to get a statistically significant sample size of early Golden Age books, the info provided from the site isn't granular enough to be of much use.

 

Still, very cool site. Majors kudos for setting it up.

 

From my POV I wasn't so interested in the GA, since we know these books are all pretty scarce. But if 2/3rds of books slabbed are 6.0 or less, doesn't this show that 2/3rds of books available are also probably going to be 6.0 or less ? What would you expect the figures on Centaur's or early Action's to show ?

[Aces]

Do you think that this analysis is only valid for comics of a certain age ?

 

One area to watch out for when looking at Census stats is the 1970-74 range, especially the non-Keys. These fall outside the lower cost Modern tier, yet are not usually valuable enough in CGC 9.2-9.4 to warrant the risk.

 

For example, I could send in 3 NM copies of X-Men 108 for the same price as 1 Captain America 180, Fantastic Four 121, or Captain Marvel 33, along with tons of other "personal faves" I'd love to own CGC'd. I have quite a few NM 20-cent Marvels (my favorite era) that would be down at CGC, if not for the ultra-high grading costs.

 

It doesn't take a genius to realize which books will be more highly-submitted, and which ones will have artificially lower Census numbers compared to their raw NM supply.

 

I think modern submissions should include all 1970s books that have a Overstreet NM value of $25 or less!

 

[CentaurMan]

But if 2/3rds of books slabbed are 6.0 or less, doesn't this show that 2/3rds of books available are also probably going to be 6.0 or less ? What would you expect the figures on Centaur's or early Action's to show ?

 

Actually, with books that old and rare, I'm curious as to how many come in at 2.0 vs. 4.0 vs. 6.0. I think the range from 5.0 to 6.0 is the absolute top of the scale for anything non-pedigree, and thus additional granularity in those middle grades would be of real interest.

 

If you get a minute, pull up a chart of all Centaurs from 1937-1939, and you'll notice how the bulk of the info you'd want to see as a collector is lost. It's like having a thermometer with 3 degrees: 0, 50, and 100. Sure, you can kind of tell someone what the temperature is, but only very broadly.

 

Again, I'm not complaining. The site wasn't meant for early Golden Age collectors, and I totally understand that. I'm just pointing out that at the extremes, even if sample size weren't a factor, the data interface would be.

[JohnBull]

Yes, I see your point ! Actually, that must be a fairly minor tweak (says the man who doesn't have to do it !). I can see why for most books the upper range was chosen, but it may well be possible to have an option box at the end of this page on whether you'd like to see the 0.0 to 6.0 range. Do specific titles, rather than publishers as a whole, go below 6.0 - I haven't time to recheck ?

[CentaurMan]

Do specific titles, rather than publishers as a whole, go below 6.0 - I haven't time to recheck ?

 

No, however you define your universe of interest (year, publisher, title, issue), everything from 6.0 and under is aggregated together.

 

You want to see a real funny histograph, pull up a specific Centaur title (say, Keen Detective Funnies) and see how all issues stack up. You know you're dealing with rare books when your sample size becomes insignificant at the TITLE level, not the issue level.

 

As an aside, this is where it would be most useful for CGC to include pedigree info in their survey. I can see that there's a 9.0 copy of Detective Picture Stories #5 from 1937, for example, and I'm really curious if that's Church, Larson, or just a random well-preserved comic. If it's not Church, then I also know that there's likely another 9.0 out there, unslabbed, from the Mile High collection, which increases my knowledge base on high grade scarcity of these early issues.

 

Another example: Right now, if the highest grade of an issue is 6.5, I see that 6.5 and think to myself that there's a hidden 9.0 somewhere because the Mile High owner has chosen not to slab his book. If the CGC survey were to say that actually that 6.5 is the Church copy, then I have a really good sense as to just how rare the highest grades will be.

[joe_collector]

I think modern submissions should include all 1970s books that have a Overstreet NM value of $25 or less!

 

I think you mean $200, since that's the current cut-off on Modern.

 

[Aces]

I think modern submissions should include all 1970s books that have a Overstreet NM value of $25 or less!

 

I think you mean $200, since that's the current cut-off on Modern.

 

 

That's fair market value of the book after it's slabbed.

I mean those $10-$25 NM guide books that become $45-$75 books 9.4 slabbed.

 

It's not worth $25 to slab them.

If they come back 9.2...

 

I think it's too hard to determine FMV on post 1970 books because the collectible bronze-modern market is still evolving. New keys still pop up (ASM 149), and others die out.

 

Silver and Gold are set.

You think FF 48 is ever going to die out?

Or Strange Tales 104 (Paste-Pot-Pete) is ever going to take off.

 

The grading price teirs should be:

All books 1970 up with a OS NM 9.4 guide value of $25 or less = $15 (modern teir)

All other bronze/moderns go economy = $29

 

Maybe the cut off point should be OS $30 or OS $40, but the system would work better wouldn't it?

 

edit:

I hate trying to figure out FMV on Bronze books when I'm not sure exactly what mutiple of guide 9.4 or 9.6 pulls.

 

Make it simple!

[JohnBull]

Another example: Right now, if the highest grade of an issue is 6.5, I see that 6.5 and think to myself that there's a hidden 9.0 somewhere because the Mile High owner has chosen not to slab his book. If the CGC survey were to say that actually that 6.5 is the Church copy, then I have a really good sense as to just how rare the highest grades will be.

 

Interesting stuff. In general, what sort of percentage of this material do you feel is slabbed ? I guess this equates to what percentage is tied up in long term collections where the owner doesn't need to slab and what percantage is circulating on the open market. Given the scarcity and the likely results (outside of a restoration check), I guess a lot is left raw ?

[valiantman]

The 2nd post (mine) in this topic on page 1 has been updated with my replies to the original post.

 

CentaurMan (and others)...

I've added a new option to "show all grades 0.5 to 10.0" which expands the table.

It makes the screen wider, so the default is still 6.5 to 10.0 with 0.5 to 6.0 aggregated.

 

I'm sure collectors won't mind scrolling left and right for the complete numbers of their favorites.

[JohnBull]

Many thanks for the answer which is more detailed than I had any right to hope for. I'm surprised by the confidence interval numbers, which were somewhat higher than I imagined.

 

I'm not sure I got the gist over of the refinement I was after. When I go to you FF 131-135 link, I can see that the number of 9.6's in that batch comes out to 27.5%, with a tabular breakdown above showing 5, 2, 4, 4 and 4 per issue respectively. I had mastered this part of the program, but what I'd actually like to see (within certain grades) is the breakdown that shows the percentage of 9.6's per issue at 23.8%, 16.7%, 33.3%, 30.8% and 36.4% respectively.

 

I know that it would be nice for the census to be much larger to give more accurate results, but by comparing like to like it does yield points of interest. The number of subs for 132 to 135 is remarkably similar (I yanked the issue/title out of thin air and I have no idea why 131 has been submitted nearly twice as many times), but the percentage for 132 is vastly different from that of 133 to 135, which are all quite similar. In this example, the number '2' as opposed to '4' is also a giveaway, but I imagine that in some cases the interpretation might not be so straight forward.

 

OK, thanks again - Gary

 

10.18pm - I appreciate I can get all the info on an issue by issue basis, but I wanted to hit a few runs that are fairly large, so it's just a time saver.

[CentaurMan]

Interesting stuff. In general, what sort of percentage of this material do you feel is slabbed ? I guess this equates to what percentage is tied up in long term collections where the owner doesn't need to slab and what percantage is circulating on the open market. Given the scarcity and the likely results (outside of a restoration check), I guess a lot is left raw ?

 

That's what I'm trying to figure out. I truly believe that fewer than 20 copies exist of some of the very earliest Centaur books, but what's complicating the matter is that Overstreet is way too low on these issues to draw them out of hiding. (Later Centaurs, from 1940-1941, are much more commonly found).

 

For example, I have a copy of Keen Detective Funnies #8, which is the very first issue. Now this took me FOREVER to find. And there's a 4" tear across the cover. There is only 1 copy in the CGC survey, and it's in 0.5 Poor condition (unrestored ). How much would Overstreet have to list this comic for before I sell it? Certainly, more than the current guide price of <$200 in Good.

 

The next 3 issues in the series haven't yet shown up in the census at all. And Overstreet lists them for <$100 in Good (I'm doing this by memory, but I think that's about right). More power to you in finding someone (who knows pre-Timely comics) who'll sell you one of these virtually non-existant books at Guide.

 

So my point is, either the earliest Centaur comics barely exist or they're tied up in private collections. And if they're tied up in private collections, current Overstreet pricing is NOT going to lure them out, because they're too @#$%ing hard to find to begin with to be worth selling for $100. But I would imagine that most Centaur collectors at this point are not going to slab the books, because the comics are in 3.0-5.0 range at best and the owners aren't selling.

 

One of these days I'll start a "Have a Cigar" thread for Centaur books. Some of the interior artwork by Cole, Everett, Burgos, Kane, Siegel, Schuster, Eisner, Kelly, Gustavson, McKay, and Guardineer is really stunning, and is never seen.

[CentaurMan]

You're the MAN, Valiantman!!! I had no idea that all I had to do was ask.

 

[valiantman]

...shows the percentage of 9.6's per issue at 23.8%, 16.7%, 33.3%, 30.8% and 36.4% respectively.

No problem.

[JohnBull]

For example, I have a copy of Keen Detective Funnies #8, which is the very first issue. Now this took me FOREVER to find. And there's a 4" tear across the cover. There is only 1 copy in the CGC survey, and it's in 0.5 Poor condition (unrestored ). How much would Overstreet have to list this comic for before I sell it? Certainly, more than the current guide price of <$200 in Good.

 

The next 3 issues in the series haven't yet shown up in the census at all. And Overstreet lists them for <$100 in Good (I'm doing this by memory, but I think that's about right). More power to you in finding someone (who knows pre-Timely comics) who'll sell you one of these virtually non-existant books at Guide.

 

So my point is, either the earliest Centaur comics barely exist or they're tied up in private collections. And if they're tied up in private collections, current Overstreet pricing is NOT going to lure them out, because they're too @#$%ing hard to find to begin with to be worth selling for $100. But I would imagine that most Centaur collectors at this point are not going to slab the books, because the comics are in 3.0-5.0 range at best and the owners aren't selling.

 

Hmm. Let's say your Keen Detective Funnies guided at $2000 rather than $200. Would you sell ? Sounds a bit of a Catch-22 if the only way you can ever get to find & buy the books is to firstly dramatically increase their value, meaning that you either i) can't afford to buy them or ii) can't afford not to sell them !

 

Please start the cigar thread. Like most of us on these boards I've seen practically nothing of these books and I'd be fascinated to learn. Don't worry about condition, we're not expecting 9.4 'books of the day'.