Found this link on the web today. It’s a model that helps you decide which diamond to buy, and weather you are paying too much. It takes into consideration all the various C’s. Very cool, and seems very accurate.
Wow, so two identical stones, one from GIA the other from AGS and the AGS is WAY cheaper than the GIA and the GIA is underpriced and a better stone than the AGS.
We are excited to see that there has been so much interest in the model on our site.
As it states on the model page, this is a Beta version, still we are rather happy with it as it has a very high ‘goodness of fit’ (R-Square, and F values etc.), and is much more accurate than other ‘models’ we have seen being used. Unlike Rapaport for instance which we understand is a frequently sighted tool in Diamond pricing, our model takes into account more than just the 3 variables Carrat, Clarity, Color. This is what makes our model more robust.
As many of those posting seem to be ‘diamond experts’, if anyone has a better model they would like to share we would be happy to evaluate it comparatively. We are doing this as an interesting side project to show how statistics and modeling can be applied to different types of problems. We feel this tool could be very useful especially for those who are not professional buyers/sellers and are not familiar with how the various diamond attributes/variables relate to each other. We do not have a vested interest in the diamond market, but rather are in the research/statistics business. As such we have nothing to gain or hide and would be happy to incorporate your feedback into future models.
Here's one way, Paul. It seems to dislike fluorescence more than it dislikes the AGS.
David,
We understand from professionals we have spoken too that the average consumer normally can not tell a glass diamond from a real one, never mind a really good one from a bad one. Obviously no model can take the place of the trained eyes of a professional looking at a diamond through a loop or under a projecting microscope. Instead effort is intended to give the average consumer another tool in order to feel more comfortable making a purchase which for most is the only diamond purchase they will make in their life, and which represents a considerable financial investment. Obviously if a collector is purchasing a classic Rolls-Royce or Ferrari, www.Edmunds.com or www.KellyBlueBook.com is of limited (though some) value. For the average car purchase though, it gives those of us who are savvy enough to locate the information on the internet, some additional knowledge. This is the intent behind this effort, a screening tool as you say, and your input is appreciated.
[/quote]
Researcher,
I understand the gestalt of your project but I think you are hoping for a much more direct, and much more static relationship between a few attributes than really exists. Saying that it’s more useful than Rapaport is debatable and, in any case, isn’t much of a compliment. It’s also better than rolling dice to decide a value. You’ve ignored the whole cut grading issue entirely. You’ll see a modifier of at least 50% based on that attribute alone. You also ignore market conditions entirely. Stores that offer better service, better trade in programs, better warranties and better payment terms and the like can charge more. They can charge more because it’s worth more in that environment and with those added attributes. We’re talking upwards of 50%. Country of origin can be an extra 15% or more. Some people actually LIKE fluorescence but even ignoring that, it’s not a straight modifier. On lower colors it’s usually counted as desirable and on higher colors it’s counted as detrimental. As has been noted above, certain lab opinions actually sell at a premium over GIA.
If this is intended to be a tool for shoppers to decide if they are getting an appropriate deal, you are offering a disservice for at least 95% of the shoppers because they value these other attributes. That’s why they shop where they shop. This gets it well beyond your example of being useless for people wishing to buy a Ferrari and is closer to being only useful only for people wishing to know the value for scrap cars when purchased by the pound.
Dhog, no offense to consumers intended. I’m sure there are many out there who have received compliments on their ‘diamond earings’ which were fake who understand what I meant. Under many types of common indoor lighting (like fluorescent) at least, it is difficult for many people to tell the difference between a good fake and a real VS1.
Perhaps we could get an opinion on the following diamonds. If you were to give these diamonds, picked randomly from online retailers, a cursory glance could you pick 2 that you might think may be ‘under valued’, 2 that would be valued just right, and 2 that would be over valued? Curious
just to make you feel good in a make believe world
Richard Sherwood, Thank you for your useful comments, we will certainly take those into consideration.
Perry, Yes we realize those on this forum may not be the most enthusiastic about this tool, yet we believe expert feedback is critical even if we are not targeting the best 5% of diamonds as you say. At this stage it is intended for the average person who does plan to purchase from/or at least benchmark against internet diamonds.
FYI,
Thank you all for your comments over the past couple days.
NOTE: A few other points about the model. It is intended to calculate round brilliant only as we understood this is the most popular shape preferred by three quarters of consumers. We do not yet allow you to analyze diamonds above 5.99 carat. This was mainly because we did not have a large sample of diamonds to look at above the 6.0 carat size, and also because we understand that the average carat size in the US, while larger in some areas like NYC, California, Boston etc., tend to be around 0.4-0.5 carat (and smaller in Europe). However, while we did have less sample even in the 5-6 carat range, price seems to be predicted fairly accurately here as well. While the underlying model is obviously predicting expected price, and we originally planned to show this, we decided that because no model is 100% accurate, and while most of our clients/larger companies would be delighted with a model as accurate as this one to predict various business outcomes, for the average layman the emotional connection with a single diamond of $x value does not allow for even <10% uncertainty. Therefore the tool was adjusted, and set up for comparison purposes to give an indication of relative value of key variables.
There are many applications for a tool such as this one. For instance the underlying pricing formulas could potentially be feedback into a database with thousands of diamonds and be used to score the gap between retail asking price and expected retail price, identifying opportunities. However, as mentioned earlier, we simply intend to test the use of multivariate analysis on something which David Atlas above refers to as “a meaningful metric for defining beauty”. The fact that “laboratories actually grade color and clarity [in different ways]” does not mean that a pricing engine based on market price is not possible, though it obviously explains why the model fit is not even higher than what we currently see.
We hope the explanation above is helpful. We look forward to serious constructive feedback from you either on how to improve our model further, or to discuss other interesting pricing techniques you may be using?
300x240.png)
