DIY??? How good (or bad) is GIA data for HCA analysis?

[Regular Guy]

Really, this is just a question about understanding the real impact of rounding.

It presumes you believe, at the outset, that proportions, in the end, will be a high predictor of performance (and the way people here talk with confidence about sweet spots in numbers...it''s clear many readers on Pricescope are converted).

The question should be answerable either purely from the point of view of understanding the mathematical analysis associated with the reporting of rounding done by GIA, and then tying that to the sensitivity HCA brings forward. Or, for those in the business, an observational answer is just as valid, or may even improve on the constraints of the question.

Basically, if you want to apply HCA analysis to the data on a new GIA cert...this is exactly and only what is done by the "search by cut" database here...how reliable is the application?

You could also say this is a question about time & money. AGS options present a premium. Not only do they not round, but they do direct analysis, and double your check, so restricting choices to only newer AGS documented options is not wacky.

But...if you want to broaden your choices...and decide to not depend only on vendors who have done the work for you, like Jonathan at GOG or Nice Ice, to name two...how effectively can you make the call?

Framing the question best takes a better skill than I''d like to bring here. One way to pose it....If you are targeting options scoring 0 - 2...how likely is it that it will really be 0 - 2, if it says it is? There may be better ways to frame the question, but hopefully this presents a beginning.

 

[JulieN]

(60T, 57D)

for 34.5/40.8, all combos in this set fall under 2

borderline steep/deep combo of 35/41: HCA=2.1
steepest 35.2/41: 2.7
most shallow 34.8/40.9: HCA=1.6
>>2 is acheived with 34.8/41
(for every GIA rounded C/P combo, there are 2 possible PAs and 5 possible CA, so 10 combos)
34.9-35.1 are 2 and under with 34.0 PA.
5 combos are 2 and under. HCA above 2 for 50%

34.5/41.2: 3.1
steepest 34.7/41.2: 3.2
most shallow 34.3/41.1: 2.3
HCA rejects for all combos

34/41.2: 2.8
34.2/41.2: 2.9
33.8/41.1: 1.9
33.9/41.1: 2
HCA>2 80% of combos

35/41 is probably the most interesting. If we say up to 2.5 (slightly leaky steep/deeps that can still be very nice,) only the steepest combo is rejected (10% rejection)

 

[oldminer]

There is a machine error present in each direct measure used for predicting cut quality. The more of these measures one uses to make their prediction, the further away from accurate are the results. Now, add on "rounding" and you have even made the reliability worse.... I don''t believe anyone in the know has an opposite opinion.

 

[Garry H (Cut Nut)]

Since HCA is a screening tool, you could screen to 3.0 and assume there may be some possible candidates included that you would miss if you screened to 2 -0.

I think it would be fair to say there can be nice HCA 2.5 stones that are slightly steep / deep.

 

[strmrdr]

For stones in the middle of the range its ok for stones at the edges of the range its not as good.
What I do is look at the extremes it could be with the rounding and see where it falls.
The pavilion angle has the greatest effect on hca score and it has the least rounding.

An aset or IS image takes off where the HCA leaves off and answers the questions.

For me the GIA numbers and a red scope image or ASET image is enough info to be comfortable with a stone.
Not that I don''t love and like more info but that would do.
Same for an AGS0 id still want the image.
AGS uses sarin data any way and is not what id consider precise.
My biggest disappointment with the AGS system is that they don''t use the helium scanners they have for all stones.
When making and using models from the data that''s like buying a corvette an putting in a 4 cylinder engine.

 

[Carlotta]

Date: 7/6/2006 7:27:46 AM
Author: oldminer
There is a machine error present in each direct measure used for predicting cut quality. The more of these measures one uses to make their prediction, the further away from accurate are the results. Now, add on ''rounding'' and you have even made the reliability worse.... I don''t believe anyone in the know has an opposite opinion.

Oldminer, what is the most accurate machine/technique for measuring a stone???? (in your opinion, LOL, I know there are many opinions)

And whatever it is, why doesn''t everyone use it???

STRM, are you suggesting that it is the Helium???? If so, why don''t more people provide that information??

 

[strmrdr]

Date: 7/6/2006 10:02:50 AM
Author: Carlotta

Date: 7/6/2006 7:27:46 AM
Author: oldminer
There is a machine error present in each direct measure used for predicting cut quality. The more of these measures one uses to make their prediction, the further away from accurate are the results. Now, add on ''rounding'' and you have even made the reliability worse.... I don''t believe anyone in the know has an opposite opinion.

Oldminer, what is the most accurate machine/technique for measuring a stone???? (in your opinion, LOL, I know there are many opinions)

And whatever it is, why doesn''t everyone use it???

STRM, are you suggesting that it is the Helium???? If so, why don''t more people provide that information??

yes helium, it isnt being marketed heavily in the US and AGS has said they dont use it because the sarin is faster and when doing thousands of diamonds the added time adds up in a hurry.

 

[Regular Guy]

A few follow up thoughts...

a) in a separate thread, Julie added:

"(what you'd really want to do is find all GIA combos of 2 (or 1.8-2.2) and test all other combinations and see what percentage of the time it goves over 2 or 2.5. I think once you got to 1.7 or something they were pretty much ok)"

It's funny....despite being accused of an analytic bent, I have never found it easy, nor have I been motivated...to learn these combos. The HCA, it seems, has been crafted for guys like me at heart, that can handle a simple construct like: 0 - 2, and like that, rather than becoming a student of CA/PA ranges.

And so, consistent with logics employed here for complementary, though possibly different reasons...and in line with what you're saying above, I wonder if calculations below your stated 1.7 or so...maybe even 1.4?....should be reserved as "sieves" though which any contemporary GIA established set of angles should be measured against?

and then...

Date: 7/6/2006 8:30:30 AM
Author: Garry H (Cut Nut)
Since HCA is a screening tool, you could screen to 3.0 and assume there may be some possible candidates included that you would miss if you screened to 2 -0.

I think it would be fair to say there can be nice HCA 2.5 stones that are slightly steep / deep.

b) repeating myself from times before, despite the intention of the "build" of the HCA as a screening tool, and while acknowledging that the author does not make a bundle on the contraption, I think that the many practical applications for HCA is by readers here who shop on the internet, and are more inductive in their shopping than deductive, unable to send back options dozens of times, and will find the application of the HCA to be helpful, functionally, for picking the option they'll begin to review, and more often then not, conclude their review with, being unable to hold onto that option at any length of time until a new one is sent to them so that they can compare the two. That ain't happening. This is, by the way, entirely Garry's fault.

c) So, what do you think? Will it be reasonable to have one standard HCA cut-off score for new GIA measured documents, low enough (like 1.4...or what other number would be conservative, but not too?) to minimize actuals over 2...and a different one for AGS options, with the understanding that -- rather than widening the funnel, we'll narrow it to only more critically let the likely good ones remain?

Maybe it depends on the budget, and generally available range of options, based on what ones looking for?

 

[JulieN]

I think a buyer would be unnecessarily limiting himself if we establish a GIA HCA cutoff.

For example, the 60T 57D 35C 41P was a 2.1. Strictly used, it would be rejected. However, there's a 50/50 chance that it would be under 2. So I think if HCA is strictly used (anything under 2) the chances would be greater than 50% it IS under 2 and the chance that it is greater than 2.5 is very small. It gets trickier if the rejection is a 2.5

 

[aljdewey]

Date: 7/6/2006 5:29:03 PM
Author: Regular Guy

b) repeating myself from times before, despite the intention of the ''build'' of the HCA as a screening tool, and while acknowledging that the author does not make a bundle on the contraption, I think that the many practical applications for HCA is by readers here who shop on the internet, and are more inductive in their shopping than deductive, unable to send back options dozens of times, and will find the application of the HCA to be helpful, functionally, for picking the option they''ll begin to review, and more often then not, conclude their review with, being unable to hold onto that option at any length of time until a new one is sent to them so that they can compare the two. That ain''t happening. This is, by the way, entirely Garry''s fault.

Ira....correct me if I''m wrong. I think you''re suggesting that the HCA should be modified to compensate for the fact that some folks use it as a de-facto selection tool instead of using it as an elimination tool as directed by the creator?

If so, I gotta disagree.

I get what you''re saying....that the restriction of economics prevents folks from shipping tons of stones back and forth, but that''s a by-product of buying long-distance, not a flaw in the HCA.

Frankly, I don''t see the need to modify the HCA at all.....despite the GIA grading reports. Why should it compensate for GIA''s choice (poor choice, in my opinion) to round key numbers? And why should it become an issue now? For quite a while, some stones didn''t have any cut data. Some only offered percentages, and using those percentages (rather than angles) came with the caveat that such rounded numbers may affect the tightness of the results. I don''t really see any difference---it''s the same caveat, but it''s just GIA rounding the numbers instead of someone else using percentages instead of angles.

 

[JulieN]

another 50/50:
61% depth, 53% table, 34° crown angle, 41.2° pavilion angle HCA 3
61% depth, 53% table, 33.8° crown angle, 41.1° pavilion angle 2
all combos of 41.1 are under 2.5
all combos of 41.2 are above 2.5

ahah.
62% depth, 57% table, 34.5° crown angle, 41° pavilion angle HCA 2
only 34.7 makes it >2 (it''s 2.1) which would still be quite nice. 10% HCA rejection

HCA is not that sensitive to GIA rounding. Keep it under 2, and the chance it is higher than 2.5 (where we are going to say you''re jumping off the cliff) is ....well, I haven''t yet found such a combo.

 

[strmrdr]

Date: 7/6/2006 6:15:05 PM
Author: JulieN
another 50/50:
61% depth, 53% table, 34° crown angle, 41.2° pavilion angle HCA 3
61% depth, 53% table, 33.8° crown angle, 41.1° pavilion angle 2
all combos of 41.1 are under 2.5
all combos of 41.2 are above 2.5

ahah.
62% depth, 57% table, 34.5° crown angle, 41° pavilion angle HCA 2
only 34.7 makes it >2 (it''s 2.1) which would still be quite nice. 10% HCA rejection

HCA is not that sensitive to GIA rounding. Keep it under 2, and the chance it is higher than 2.5 (where we are going to say you''re jumping off the cliff) is ....well, I haven''t yet found such a combo.

That the hca is overly harsh on stones with low 41 degree pavilions is well known.
That it is a little soft on shallow/shallow combos is also well known.

The thing is that it cant judge optical symmetry and tightness which make a large difference when you start playing at the edges.

 

[Garry H (Cut Nut)]

Date: 7/6/2006 10:52:30 PM
Author: strmrdr

The thing is that it cant judge optical symmetry and tightness which make a large difference when you start playing at the edges.

The explanation here - http://diamonds.pricescope.com/ideal.asp is attached below: some bits are highlighted blue here to reinforce Storms comments

The ''X'' on the chart below marks the pavilion° / crown° position on the closest table size chart. The white outline shows the AGS ideal ‘candidate’ range and the green dotted line shows the GIA Excellent ‘candidate’ range. A conservative (but more costly) choice would be a diamond with proportions in the overlap of both AGS and GIA ideal and excellent ranges.

Holloway Cut Adviser was launched on Pricescope in 2001 after 15 years study and lecturing diamond cut. I developed HCA with the Firescope and DiamCalc.

HCA is only for round diamonds with known Table, Crown, Pavilion and Total Depth inputs; it has no idea about the diamonds symmetry, poor minor facet proportions or various cutters tricks.

Use it to reject known poor performers and narrow down your selection. Please do not use it for final selection; it will not replace an analysis with an Ideal-scope and/or an independent appraisal.

Stones listed here are ranked by HCA scores, and many have ideal-scope photo''s and other cool stuff.

GIA''s new system is similar to HCA but with symmetry, polish and minor facet data, but the data on GIA reports has been severely rounded. In order of importance: Pavilion 0.2°, Crown 0,5° lower girdle depth 5%. Steeper crown and deeper pavilion GIA graded ''Excellent'' diamonds have a dark ring in the table area and will show the dirt on the pavilion.

Diamonds that rate below 2 (red on the chart) are unlikely to you too much leakage darkness, overly thin or thick girdles and fish-eyes. But there are other negatives that HCA can''t predict. Many people who can focus closer up than 10 inches (25cm) prefer diamonds that ar marked in the upper zone marked on the chart below.

Figure. This chart indicates some diamond proportions and information that you may find useful.

Date: 7/6/2006 10:52:30 PM
Author: strmrdr

That it is a little soft on shallow/shallow combos is also well known.

Shallow stones (lower left on the chart) look darker if you have excellent close up vision because your head obstructs more light sources which makes a shallow diamond appear darker. But shallow diamonds have a bigger spread, and are great for pendants and earrings, where normal social viewing distances apply.

Date: 7/6/2006 10:52:30 PM
Author: strmrdr

optical symmetry and tightness which make a large difference when you start playing at the edges.

Stones near the center of the red region (the lowest scores) are least affected by symmetry variations. Alternatively hearts and arrows diamonds, which have excellent optical symmetry, but often HCA scores around 2, may out-perform diamonds with lesser symmetry and lower HCA scores.

Deeper stones in the upper green and blue zones have more leakage and often appear dark just inside the table and the outer girdle edges; they are best set in open backed rings so light can get in the bottom or pavilion. But small ''vee'' shaped leakage zones near the outer edges of a diamond, as seen with an Ideal-Scope, can add to a diamonds contrast and brilliance.

Trivia

HCA caused a bit of a stir when we first offered it free on Pricescope in 2001; like any new concept that affects diamond sales, HCA was not liked by many vendors. GIA spent millions on cut research, Octonus and Moscow State University used their brains, and AGS did a seat of the pants study like HCA. All support the inverse Crown and Pavilion theory.

HCA is fine-tuned from time to time - results may change without notice. Remember - only use HCA to reject diamonds & narrow down your selection. A full technical description of HCA can be found at www.diamond-cut.com.au

Garry Holloway, FGAA DDT