shape
carat
color
clarity

Diamond Shopping Strategies revisited, Data-mining, and Expa

Discussion in 'RockyTalky' started by Regular Guy, Apr 15, 2005.

Thread Status:
Not open for further replies. Please create a new topic or request for this thread to be opened.
  1. Regular Guy
    Ideal_Rock

    Messages:
    5,951
    Joined:
    Jul 6, 2004
    by Regular Guy » Apr 15, 2005
    OK, so this post is about investigating an idea about how to expand your options in the diamond selection process….are you game?

    For those willing to help, I also have 3 questions, bolded, at the end.  More immediately below, some ideas to share…

    Anyone there a fan, like me, of the show that has become Boston Legal?

    It had been called: The Practice (and by the way, do you know if they’ve moved that inferior new doctors show into it’s old time slot only temporarily…now that’s what I really want to know!).  And, when it was “The Practice,” frequently they worked Plan B, where when they would go to someone’s defense, they would begin by going for Plan A, whatever that was…and if that turned sour, they would go for “Plan B.”

    In my oft mentioned strategies for shopping, I mention a Plan B.  (Again, btw, both John & Neil have acknowledged reading this…anyone else have any feedback on it in principle?).  For awhile, I’ve been interested in another part of the Pricescope database, and I’ve finally thought of a way to integrate it as a shopping resource.  If others have done this already, do tell.

    This one is the Price Stats database, under Prices at the top on Pricescope.  It’s designed to replicate the type of content you get with the Rap sheet frequently mentioned here, but sort of do it more meaningfully, because it references real prices….and when you go to find out the market basket cost for a given diamond (I have been testing size .9 - .99, G VS2), you can find out prices based on actual prices and those supposedly availability with vendors, today.

    So, let’s review…the proposed idea is this:

    For Plan A….always, go to the search by cut database, and find what you want among those options Pricescope vendors have already thought to bring in house.  But, if the item is either unavailable, or more likely, you want to see if, by fishing more broadly, you can do pretty well, and pay less, with something similar….with the existing Plan B….you go to the Quick search, entering in the diamond’s characteristics you’re looking for, and then check either or both AGS0 and/or H&A (which also checks AGS0).

    But, I’ve recently thought about integrating the Price Stats section as a way to expand the options in plan B.  First I messed with the output of AGA charts H&A through AGA 3B, saw the general pattern of real decreasing prices, going from higher to lower quality (with one or two exceptions), and also tied the changes to the expected changes in table & depth, which is all I could figure Leonid is capturing in an algorithm to make these charts work.

    Looking at the output you’d want to predict for 1A, you might expect there’d be less options as compared to the AGS0 sort in the Quick Search, since we understand Dave’s 1A is designed to be a more narrow swath.  Doing a comparison on just the one grade, though, I see there’s 3 times as many.  If you consider that when searching for AGS0 in the Quick Search, Leonid is looking for text in the comments line that actually says “AGS0,” or something resembling that, it’s not unreasonable to assume that he is pulling a more narrow sort with this approach, especially understanding that depth & table data do not make an AGS0 alone.  And so, this outcome is not unreasonable.

    But one principle question emerges for me:

    Why aren’t any of the picks in the Quick Search scored for AGS0 the same as those found in the sort of those fitting 1A?  There should be some crossover, but I saw none, at least when looking at the several options displayed from DCD.

    Maybe Leonid can either help clarify how options get into those that come up for the Price Stats search, or maybe he can clarify this non-overlap question directly?

    Also two caveat/notes: I also checked to see if there was any real world relevance to any of this.  For example, are the diamonds that show up in these databases even showing in the databases of the vendors.  Looking at my target option, for the 3 diamonds I saw appearing from DCD, I almost gave up, but did see one was a match, and it was one in the AGA charts.  I attribute the missing ones to time delays, and their at least possibly having been sold, since posted in Pricescope.  The other problem re-emerged, however…and that is seeing that how Pricescope categorizes options may not match how vendors do their matching…and this is a problem Superidealist (where have they been?) pointed to not long ago. The AGA 1A match came up in the list that DCD categorized only as Premium (versus those found in their ideal sort).  Presumably, Jim has no more info than Leonid, so apparently their algorithms for sorting are different, and sometimes the disparity is more wide as well.  This, a problem for another time, since the idea is to expand options, rather than contract them (except to note that the consumer, if shopping for something like ideal, could do well to keep looking until the respective databases seem to match on that info).

    Two more questions:

    Do others see this as a promising strategy for expanding other options to consider for searching?

    Moreover, to any vendors/jewlers out there who use the non-Pricescope version of what is presented here as the Quicksearch….is there a set of protocols you use you’d like to share to optimize for your prospects, to get them the diamond they’re looking for, when they’re working with a somewhat tight budget.  I have to think your assortment of databases must look a lot like the ones that are here on Pricescope.  Wink recently described -- having looked at some 10 options that he had little data about, and he took the time to comment on -- which one he would think is the most promising.  Anyway, some ideas about how the professionals here look to serve their clients would be of interest as well.

    Thanks for your interest, and thoughts.

    Best of wishes!
     
    


    


  2. lostdog
    Shiny_Rock

    Messages:
    179
    Joined:
    Dec 14, 2004
    by lostdog » Apr 15, 2005
    Ira,

    I am just not following what you are getting at here with the two plans.

    Can you sort of summarize it a few sentences?
     
  3. Regular Guy
    Ideal_Rock

    Messages:
    5,951
    Joined:
    Jul 6, 2004
    by Regular Guy » Apr 15, 2005
    Good morning lostdog...with a fair amount of readers, probably many have thought the same.

    Can you help me work out a set of steps for shopping, is what I''m after. I had earlier suggested 2 steps (see right here below)...and now I''m suggesting one more step to increase your chances of finding a winner.

    Let''s say you''re shopping for a diamond, are attracted to any type...say size of about .9 carats, color G, clarity VS2, and you want it to be well cut. You don''t want to mortgage your house, etc.

    First stop, the search by cut database, where a quick sort will find you 2 lovely options from GOG, costing either $5387 or $5427. Maybe I wanted to actually spend under $5K, though...what to do?

    Next stop, the quick search, input the same data, specify (check the box) AGS0 to pick, and find the same 2 coming up, plus 5 more to choose from, and so now you''ve got 7 options to look at, all called "cherry" by somebody, ranging in cost from now $4074 to $5427.

    Can you retrieve more to consider? That''s what this post is designed to be about. Let''s go to the well one more time.

    Under "Prices" up top, go to the drop down, and select the "Price Stats" option. Select your carat range and generate. Find the prices at average and low for your size clarity & color, and select the linkable option. There you''re taken to a page of options that are designed to in some way describe themselves as -- in this case -- AGA 1A, .9, G, VS 2, and voila, you have 20 more to pick from, ranging in cost from $4074 to $4822, and you''ve tripled the number of diamonds you have to consider!

    The goodness of the data is always something to consider, and I''ve included that caveat...but if this additional strategy leads you to find two or three options you''d like to consider more closely, where before you had either 0 or 1, you can then work with your vendor to look at these, get their opinion, and only then, have them called in to evaluate, and buy.

    So, do you or does anyone have any thoughts on how this might help ones shopping? Personally, I might like to get a better handle on what variability of data I should be expecting, when selecting the AGA 1A category...essentially confirming that we''re working with basically depth and table info only. Even so, my understanding is that if even if that is the case, we''re working with no less than the tools any jeweler or vendor ever works with, for the most part, and so -- increasing your options this way should only be able to help.

    Oui?
     
  4. oldminer
    Ideal_Rock
    Trade

    Messages:
    6,271
    Joined:
    Sep 3, 2000
    by oldminer » Apr 15, 2005
    I had a relatively difficult time with thie above posting. I am sure that there is some merit in what is being reviewed, but I see the AGA 1A mentioned a few times. What has happened is that cutters have found ways to get AGS 0 but somehow miss the very fine Tolkowsky settings of the AGA 1A. They have done this to "play the game", not make diamonds better necessarily. The more weight they can retain, and still get the magic AGS 0 grade, the more money they can make with a good selling diamond. The AGA-1A leaves little room for compromise of cut. What actually happens is that the most commonly cut range of AGS 0 stones looks like a donut, one with a hole in the center. AGA 1A is the hole with farily few stones cut that way , but the wider parameters which fit AGS 0 are populated with many available stones.

    How Pricescope searches for these facts, I have no idea. What comes up in a search is inconsistent, but it is due to the way people have entered the details of the cut, or the lack of details of cut.

    Having alternative ways to find more possible stones sound useful.....Hope others understand the suggestions better than I do.
     
    


    


  5. Regular Guy
    Ideal_Rock

    Messages:
    5,951
    Joined:
    Jul 6, 2004
    by Regular Guy » Apr 15, 2005
    Dave,

    So, thanks for reading, and not to mention, for coming up with the cut charts.

    Unless I'm mistaken (and frankly, I'm hoping you and others will try to take a second and do see how Pricescope does come up with these facts...) what Pricescope does is add some switches behind the scenes, otherwise unavailable, allowing you to get leverage for the information here, and sort with power otherwise somewhat unavailable, with just the switches that are available on the Quicksearch.

    - Without the AGA tables built into Pricescope, and just adjusting for table & depth, you get as many as 69 options (though not a bad number [​IMG], having constrained table to 53 - 58 and depth form 58 - 62.3. Then, instead, to compare, use the AGA tables built in (follow the detailed steps below), and see this number narrowed to 20.

    - Don't be afraid to push the buttons
    - At the top (just scroll up), where it says: Home, Forum, Prices, Knowledge, etc... put your cursor on prices, and come down to Price Stats, and you'll come here.
    - Although you can play with this screena little, it's optimized for 1A right now, so just hit the generate button
    - come here
    - find the grid for what you want...and in my example, find the size of .9 - .99, find either in the left or right columns (it doesn't matter) your color or clarity choices. If G, VS, find the intersection of those combinations, and hit the link. If on the right, it will say "4334."
    - go to the presentation of 20 options, which will have narrowed your choices considerably, I now see, vs. trying to re-create the AGA 1A yourself, with just table & depth info.

    Leonid, maybe you can you help clarify what further constraining you are doing behind the scenes, here?

    Thanks again.
     
  6. denverappraiser
    Ideal_Rock
    Trade

    Messages:
    8,612
    Joined:
    Jul 21, 2004
    by denverappraiser » Apr 15, 2005

    Ira,


    For the most part, I think your strategy is valid.

    Much of the problem may be with the data that is supplied. Not every vendor has every tidbit on every stone. Pricescope apparently makes it possible for a dealer to list about 20 fields of information for each stone when they post it to the database. They can then build filters to sort the stones in a variety of ways that may be interesting or useful to users. Users are currently given the opportunity to sort based on about half of these criteria although others are obviously part of the algorithms as you have observed. It looks to me like the dealers omit at least some of the data for a significant fraction of the stones in the database. I think it’s a mistake to assume that these will therefore be inferior stones. I’m reasonably confident that this is to save data entry labor and not really an effort to conceal important details. In the case of stones that they own, I suspect it''s just a matter of time constraints. In the case of stones where the listing dealer doesn’t physical have access, they must rely on their supplier to produce this data. Some suppliers are more forthcoming than others. Moreover, I suspect that many of the best ‘deals’ are likely to occur in this category for those who are willing to sniff them out for exactly the same reasons that EGL grading reports may be an indication of a similar stone with a lower price (or maybe not).
    Neil Beaty
    GG(GIA) ISA NAJA
    Independet Appraisals in Denver
     
  7. dastewar
    Rough_Rock

    Messages:
    43
    Joined:
    Apr 27, 2005
    by dastewar » May 2, 2005
    Dear oldminer:

    Should the average consumer care if something is AGS0 but NOT AGA 1A. Also, should we care that some AGA 1A''s are "barely" in there and could be within a marginal measurement error of 1B?
     
Thread Status:
Not open for further replies. Please create a new topic or request for this thread to be opened.

Share This Page