Reliably and repeatedly grading the color of diamonds

[adamasgem]

Here is the data to chew on, but I caution that the sample sizes of AGS and EGLUSA data are much smaller than the GIA sample size, but as I said above, if you even average the AGS and EGLUSA data, something is inconsistent, and one can draw any conclusion they want to, including some serious sinister ones. The data is the data.

One might say from the data that AGS and EGLUSA are overall more conservative in color grading (as a matter of disclosure, both AGS and EGLUSA are SAS2000 clients of mine, but to my knowledge, they DO NOT USE the SAS2000 for colorless color grading, but rather for treatment detection when stones are flagged for screening).

On the flip side, one might say that GIA gets to see better color stones on the whole, but that could be a stretch of the imagination.

I wonder if I will be allowed to attend the GIA gemological symposium next year, and, if I do, whether I should wear a bullet proof vest

 

[strmrdr]

This chart I don''t find too surprising.
AGS mostly does well cut stones and most of the AGS0 stones available are not D,E if you did the numbers on whiteflashe''s GOG''s and Paul''s network Im thinking you would find that the ags curve is pretty accurate with the majority being in the f-h range and very few in the D,E range.
Im fairly certain that Paul and or Brian will confirm that the D,E rough is too pricey for the weight lose of cutting to AGS0 and the selling price required too high to move them quickly.
Therefore you wont see a lot of ags0 D,E.

That goes double if AGS is going to grade them lower than GIA and would increase that effect.
An interesting number would be the F with fluoresecence gia vs ags.
If there are a lot smaller percentage in the ags stones that would be an indicator that something is going on and support your other charts.

 

[adamasgem]

Date: 11/24/2005 1:32:59 PM
Author: strmrdr

This chart I don''t find too surprising.
AGS mostly does well cut stones and most of the AGS0 stones available are not D,E if you did the numbers on whiteflashe''s GOG''s and Paul''s network Im thinking you would find that the ags curve is pretty accurate with the majority being in the f-h range and very few in the D,E range.
Im fairly certain that Paul and or Brian will confirm that the D,E rough is too pricey for the weight lose of cutting to AGS0 and the selling price required too high to move them quickly.
Therefore you wont see a lot of ags0 D,E.

That goes double if AGS is going to grade them lower than GIA and would increase that effect.
An interesting number would be the F with fluoresecence gia vs ags.
If there are a lot smaller percentage in the ags stones that would be an indicator that something is going on and support your other charts.

Hi Strmrdr..

What is interesting is the the EGLUSA data tracks the AGS in the D-E range.. I would think that the distribution of rough woud be pretty uniform, but you could argue that AGS is more conservative so cutters send to GIA for better grades, as you say.

I''d like to see a large sample set of HRD and IGI and more data from AGS ane EGLUSA.. as well as other labs world wide to see their sample distributions. The factors of 2 in relative D and E populations would seem unusual. Interesting isn''t it?

 

[adamasgem]

Here is the same data set in bar chart form

 

[adamasgem]

In the wholesale database I looked at, the 900# Gorilla

is GIA, with about 82.5% of the paper across ALL labs for the US wholesale to retail trade (That may be decreasing given recent events) . That is not to say, that there are vastly more stones graded by labs like IGI and EGLUSA, and the smaller labs, some of whom are just doing "paper", some trying to do good work, and a lot writing pie in the sky "appraisals" for the chain stores and mass merchandisers. And there are a lot a stone offered for sale with just the seller''s representation of the grade.

It would be interesting to hack into all the labs internal databases to see what the overall statistics of their grades are... we can guess with a pretty reasonable degree of confidence that the lower clarity grades typically don''t go to the top tier labs for grading, but I don''t know anyone at CIA anymore.

 

[RockDoc]

RE: HACKING ........ Do you really need someone at the CIA? I don''t think you do anymore.......

From all the spam, viruses, harvesters of personal information, and phishers, who continuously annoy the best of us ( including Microsoft etal ) it probably isn''t too hard too find a young teen that might get this stuff for you.

( This post is intended to be humorous even though it probably is factual But it is amazing how these younguns can hack US military sites)

Have a good turkey day!

Rockdoc ( Compu-Grape)

PS: Is there one of those books.... THE IDIOTS GUIDE TO HACK IN TO SOMEONE"S DATABASE ?

 

[Rank Amateur]

Looking at the charts it seems that DEF is more common than IJK. So why the talk about the "rarity" of DEF raising the price of the colorless stones?

Interesting.

 

[JohnQuixote]

Date: 11/25/2005 9:51:02 AM
Author: Rank Amateur
Looking at the charts it seems that DEF is more common than IJK. So why the talk about the 'rarity' of DEF raising the price of the colorless stones?

Interesting.

Remember that these are only AGS/GIA/EGL stats. Diamonds that will be graded DEF by one of the 'premier' labs are sent there for that purpose, to acquire the strong grading document (in the same way that AGS receives an overbalance of fine-make rounds). Many near-colorless (and below) diamonds are sent to softer labs in hopes of receiving a higher color grade than the 'premier' labs would give them...or they are not sent for grading at all.

 

[adamasgem]

Date: 11/25/2005 11:50:09 AM
Author: JohnQuixote

Date: 11/25/2005 9:51:02 AM
Author: Rank Amateur
Looking at the charts it seems that DEF is more common than IJK. So why the talk about the ''rarity'' of DEF raising the price of the colorless stones?

Interesting.

Remember that these are only AGS/GIA/EGL stats. Diamonds that will be graded DEF by one of the ''premier'' labs are sent there for that purpose, to acquire the strong grading document (in the same way that AGS receives an overbalance of fine-make rounds). Many near-colorless (and below) diamonds are sent to softer labs in hopes of receiving a higher color grade than the ''premier'' labs would give them...or they are not sent for grading at all.

Thanks John, for responding to "Rank Amateur".

The charts speak only to the AVAILABILITY OF GRADED STONES, not to the real distribution of the all stones MOST OF WHICH are in the I1,I2,I3 and boart sold by the mass merchandisers that are NEVER really "graded", even though they may have "paper". OH, to be able to hack into some "labs" data base, or get the "rough stats" from DeBeers. THE VAST MAJORITY OF DIAMONDS RECOVERED FROM MINES, are "industrial". Look at the average price per carat of ROUGH recovered from mines world wide, and the numbers are only in the 100 to 300 dollars per carat, and cover all sizes, colors and clarity ranges.. The High colors, clarity and sizes over 1/4 carat, those that are typically graded, according to what I have allways heard and read about make up a small fraction of total diamonds recovered from mines, but that is a different question from what is being discussed here..

I guess that is why Garry tired to "interpolate" my original graph, and I maybe inappropriately chewed him up for it. I don''t like to make any assumptions about data I don''t have....

 

[diamondsbylauren]

Call me crazy- ( most of you do anyway) but isn''t using a list where the same stone is listed 5 times going to throw the entire study out of whack?

 

[adamasgem]

Date: 11/25/2005 3:05:28 PM
Author: diamondsbylauren
Call me crazy- ( most of you do anyway) but isn't using a list where the same stone is listed 5 times going to throw the entire study out of whack?

Hi Crazy.. It is not a stupid queston, but the data I presented, as well as what Leonid is doing, are pulled from wholesale databases without duplication, or combining each cutter's inhouse inventory, where there are not "duplicates" as one would find by combining "virtual inventory" consumer web sites to get the required sample sizes. Both leonid and I know what we are doing with regards to this issue.

Although a consumer might want to play with this issue, he or she would have to stick to ONE retailersr's stone list, but the sample size they got, would not necessarily give meaningful results.

That is why I suggested hacking into lab's data bases (or getting cooperation from them to release their internal statistics, which I doubt would happen in MOST cases[I am working on getting some cooperation there] ), but even there, there might be some undetectable duplicates, as cutter's sometimes "shop" for the best grade buy submitting the same stone to two or three labs.

And I specifically did not include SI3, and I1-I3's in my study, because they make up a very small (<1%) of wholesale graded liists, but make up a very large percentage of the mass merchandisers inventory and sales, and the "real" data is not available there except maybe by supeona in the right instance

 

[diamondsbylauren]

Marty- you old hacker you......hehehehe


Seriously- I don''t think that even if GIA and EGL opened up thier computer listings that we would have anywhere near what could be called a "comprehensive" veiw of the average quality of polished diamonds.
AS you point out- so many SI3-I2 stones never see a gem lab- and plenty of those are passed off as SI2''s by lesser quality sellers.
Same for K-L-M....and lower colors. There''s quite a bit of that cut which never sees a list, or a lab.

 

[JohnQuixote]

Date: 11/25/2005 7:12:11 PM
Author: diamondsbylauren
Marty- you old hacker you......hehehehe

Seriously- I don''t think that even if GIA and EGL opened up thier computer listings that we would have anywhere near what could be called a ''comprehensive'' veiw of the average quality of polished diamonds.
AS you point out- so many SI3-I2 stones never see a gem lab- and plenty of those are passed off as SI2''s by lesser quality sellers.
Same for K-L-M....and lower colors. There''s quite a bit of that cut which never sees a list, or a lab.

That''s why we must try further upstream.