FMV and GPA Price Manipulation 2 2

[theCapraAegagrus]

FMV = whatever I'm willing to pay.

[Squad008]

My best guess regarding the extreme outliers is that they were eBay BIN listings where the seller listed high but used the eBay feature to privately send an offer to an interested buyer at an undisclosed lower price. When sales complete this way, eBay shows that the item sold for it's full asking price and GPA picks up the data as though the it sold for that amount. 

If those sales weren't BIN through eBay then who knows. Regardless, I always disregard extreme high and low sales when analyzing a book. Low sales have the same issue when it comes to FMV as the book itself could have presented poorly, been damaged (gotta love the high grade sun damaged CGC books on eBay) or possibly a seller got the book super cheap and doesn't want to wait a few weeks/months to get FMV and lists it cheap to move right away.

Either way, high sales and low sales aren't manipulation, they are just outliers in a set of sales data that the individual looking at the data needs to decide whether or not to exclude. 

Oh yeah, let's not forget how many sales are made when eBay has a 10% to 15% off coupon. Yes the Seller gets full asking but the buyer paid 10-15% less and eBay subsidised the rest. Those coupons can create higher sales numbers especially on books where the discount doesn't exceed the $100 maximum discount that eBay usually limits it too.

Edited April 16, 2019 by Squad008
Added comment

[ExNihilo]

Unpopular opinion here, but I just don't get seller's stubbornness at times to abide by GPA pricing.  I've reached out to sellers who respond to me "oh, this book is $1000 on GPA and I think that's fair so I can't go lower."  But isn't GPA just completed eBay sales?  My problem is that it doesn't factor in the 10% cut eBay takes or the PayPal fees.  So that book that's sold for an average of $1000 on eBay, the seller is really only pocketing about $900.  So when I come in with an offer of like $950, I honestly don't think I'm being unreasonable.  You make more than you would have on eBay, I save more than I would have on eBay.

And if we're being honest, it's even worst at cons where some books are priced well above eBay, but they'll cut you a deal for paying cash.  Like, I might as well just go to eBay at that point and pay the lower rate, get my 1.5% eBay bucks, plus my 3-5% in credit card rewards.  The only reason to buy from an actual vendor instead of eBay is because you can see the quality of the book for yourself and you can avoid any potential shipping hassles.

[Ryandw80]

Newbie here is there a FMV or comic stock market app/website so I could keep track of my comics?

[crassus]

I realize this thread just got bumped from the dead, but it seems an appropriate place to ask my question, as the OP's topic is on my mind.

How does Go Collect calculate FMV exactly? 

I just started consulting Go Collect to try and self-appraise my collection, and I understand its best to survey as much and different data as possible. 

That said, to use two examples, Go Collect has for an FF 1 in 2.5 an FMV of 12.5k and for an FF 2 in 6.5 an FMV of $6.75k, however the list of sales to the right, given the dates of sales and comparing those sales dates to GPA, does not give figures remotely comparable to the FMV. If the last sale for the FF 2, for example, is Dec 21 2020 (Heritage) than that sale is recorded on GPA as $3240 as the Heritage sale. 

I understand that real values in any grade tier require comparison across tiers, as more recent sales in other tiers may float every boat, etc...so should I assume that the FMV figures given are projections? Or is there additional sales data only available to subscribers? At any rate I'm just not sure what method is used for these FMV quotes. 

[thunsicker]

1 hour ago, crassus said:

I understand that real values in any grade tier require comparison across tiers, as more recent sales in other tiers may float every boat, etc...so should I assume that the FMV figures given are projections? Or is there additional sales data only available to subscribers? At any rate I'm just not sure what method is used for these FMV quotes. 

I would assume there is a proprietary equation that they use.  Something with best fit lines.  When I use the 90 day averages for those comics and do a best fit exponential line to it it gives me the equation:

Value = 919*e^(.2707*grade)  which works out to around $5400 for a 6.5.

[SuperBird]

3 hours ago, Ryandw80 said:

Newbie here is there a FMV or comic stock market app/website so I could keep track of my comics?

https://comics.gpanalysis.com/analyse-prices is the go-to site. 

 

[crassus]

14 minutes ago, thunsicker said:

I would assume there is a proprietary equation that they use.  Something with best fit lines.  When I use the 90 day averages for those comics and do a best fit exponential line to it it gives me the equation:

Value = 919*e^(.2707*grade)  which works out to around $5400 for a 6.5.

Thanks for this, and the 90 day averages you reference include other grade ranges, yes? I'm just trying to figure out what the premise of the model is that gets significantly lower values into higher values for an FMV....

[thunsicker]

3 minutes ago, crassus said:

Thanks for this, and the 90 day averages you reference include other grade ranges, yes? I'm just trying to figure out what the premise of the model is that gets significantly lower values into higher values for an FMV....

Right (and I should mention that I pulled data from GPA not GoCollect since that's the one I'm subscribed to).

Here's the grades for FF #2 that have sales in the last 90 days and the 90 day averages for those grades:

8.5	9600
5	3550
4.5	3060
3	2300
2.5	1420
2	1397
1	1519

Plug those in in Excel and do a scatterplot.  Add a best fit line (exponential) and display equation on chart.  That will give you a rough idea.  Note that this method tends to be a bit weaker in higher grades but it works pretty well for the center:

[crassus]

Ok thanks for taking the time to explain, so if I understand you correctly, the FMV is a projection based on all the available sales data for a given book, yes? And this allows them to plot any given grade tier on a graph like yours above, yes? 

[thunsicker]

1 minute ago, crassus said:

Ok thanks for taking the time to explain, so if I understand you correctly, the FMV is a projection based on all the available sales data for a given book, yes? And this allows them to plot any given grade tier on a graph like yours above, yes? 

To be fair, I have no idea of how they do their model, but I would assume it is something similar using some statistical algorithms.  If they had a data scientist I'm sure they could get all kinds of interesting information with the volume of data that they must get through there.

[crassus]

8 minutes ago, thunsicker said:

To be fair, I have no idea of how they do their model, but I would assume it is something similar using some statistical algorithms.  If they had a data scientist I'm sure they could get all kinds of interesting information with the volume of data that they must get through there.

I understand, would there also be choices made between different kinds of models? So you did your own just here, and got one set of values, could you have made choices to tweek your model that could have an influence on the kinds of values returned?

[thunsicker]

6 minutes ago, crassus said:

I understand, would there also be choices made between different kinds of models? So you did your own just here, and got one set of values, could you have made choices to tweek your model that could have an influence on the kinds of values returned?

There are always choices that can change the values returned.  Use last 12 month average? Use last 90 day average? Use last sale? Account for specifically how long ago the sale was? Change the type of equation you use (Linear? Exponential? Power? Polynomial?) Account for page quality? Again with all the data that exists out there a data scientist (along with someone who 'understands' comics) could come up with a fantastic set of equations that would account for all kinds of things.

And of course the caveat of all this is that not all 9.4's or 6.0's or 2.0's are the same.  Some look much better than others and would sell for much more.

Edited April 15, 2021 by thunsicker

[crassus]

6 minutes ago, thunsicker said:

There are always choices that can change the values returned.  Use last 12 month average? Use last 90 day average? Use last sale? Account for specifically how long ago the sale was? Change the type of equation you use (Linear? Exponential? Power? Polynomial?) Account for page quality? Again with all the data that exists out there a data scientist (along with someone who 'understands' comics) could come up with a fantastic set of equations that would account for all kinds of things.

And of course the caveat of all this is that not all 9.4's or 6.0's or 2.0's are the same.  Some look much better than others and would sell for much more.

Thanks, one further question (and I appreciate your time, I am not mathematically inclined so it helps me) since this kind of analysis offered by Go Collect is all for the benefit of a growing market in collectibles, could Go Collect (conceivably or hypothetically) choose models that favoured either more conservative or more optimistic values? In other words, could a given model favour the projection of inflationary values?

[thunsicker]

1 minute ago, crassus said:

Thanks, one further question (and I appreciate your time, I am not mathematically inclined so it helps me) since this kind of analysis offered by Go Collect is all for the benefit of a growing market in collectibles, could Go Collect (conceivably or hypothetically) choose models that favoured either more conservative or more optimistic values? In other words, could a given model favour the projection of inflationary values?

I suppose they could tilt the model if they felt their membership was more interested in setting prices for comics they wanted to sell rather than finding reasonable prices for comics they were interested in buying.  I don't mean to say everyone needs a data scientist, but if they did have one you would hope that he would push the model towards giving the price that was most predictive.  But as you point out most businesses would prefer the model that makes the company the most money rather than the one that is the best predictor.

[crassus]

9 minutes ago, thunsicker said:

I suppose they could tilt the model if they felt their membership was more interested in setting prices for comics they wanted to sell rather than finding reasonable prices for comics they were interested in buying.  I don't mean to say everyone needs a data scientist, but if they did have one you would hope that he would push the model towards giving the price that was most predictive.  But as you point out most businesses would prefer the model that makes the company the most money rather than the one that is the best predictor.

Personally I give them the benefit of the doubt, ultimately you only stay in that game if your product (your analysis) stays relevant, and if you want people especially to consider forecasting based on that product than you had better be "most predictive" as you say. Its also the case that everyone has a stake in a growing market continuing to grow, and there is no avoiding inflation, but I think what's new (for me at least, being old) is how aggressive a fact inflation is now for the hobby. The OSPG is premised on a world of graduated grade tiers with fairly consistent progressions between tiers, so from G to VG is this much, and from VG to Fine is that much etc....but looking at GPA, especially with trending books, everything becomes compacted in value towards the bottom of the scale, the old simple differentials of the OSPG disappear in a flood, suddenly everything is expensive and getting pricier, no matter the grade...inflation now is such a Godzilla. 

[valiantman]

6 hours ago, crassus said:

Thanks, one further question (and I appreciate your time, I am not mathematically inclined so it helps me) since this kind of analysis offered by Go Collect is all for the benefit of a growing market in collectibles, could Go Collect (conceivably or hypothetically) choose models that favoured either more conservative or more optimistic values? In other words, could a given model favour the projection of inflationary values?

Yes, and there would be "valid" ways to do so.  If the last two sales for a book were $100 (a year ago) and $150 (six months ago), you could say that although there have been no more sales, the price "should be" $200 today.

However, a better approach is simply to state what "was" and not what "could be".  In other words, 12 month average = $125, no sales in the last 90 days, last sale $150 six months ago.  Which is what GPAnalysis does.

[valiantman]

For what it's worth, mid-grades are beginning to "settle" into a price-per-point calculation that makes sense to most people.

If a book is $5,000 for a 5.0, then it's probably $6,000 for a 6.0 and $6,500 for a 6.5.  When you get to the higher grades, it doesn't work anymore, but some of the examples described above are G to VG, VG to Fine, and those are "price-per-point" areas.

Likewise, if it's $150 for a 5.0, it's probably $195 for a 6.5.  Not as easy to calculate in your head, but exactly the same math as the $5,000 for 5.0 to $6,500 for 6.5 example.