CNG Auction Estimates - How Low Can You Go?

Discussion in 'Ancient Coins' started by John Conduitt, Aug 27, 2020.

  1. John Conduitt

    John Conduitt Well-Known Member

    As another CNG Electronic Auction passed, it was not hard to notice that most lots went for more than their estimates. In many cases a lot more. Electronic Auction 475 saw a combined hammer price of $230,500 against a pre-sale estimate of $101,880, and it’s not alone. All 19 CNG Electronic Auctions since December 2019 (when they first published their total estimates) have seen the combined hammer price exceed the total estimate by more than 91% (i.e. double). The average is 131% higher.

    Not great for those of us looking through the catalogues hoping to add to our collections. But why is this?

    It was pointed out in a recent thread that low estimates are used to generate bids (by getting more people interested) and to get us attached to particular coins (so we’ll eventually bid more to get them). Moreover, since CNG set most of their minimum bids at 60% of the estimate, a lower estimate helps ensure fewer lots go unsold

    But if an estimate is simply 166.67% of the lowest price the seller would accept for their coin, is that really an estimate? It’s certainly not a great valuation, given the average result is more than double. You wouldn’t want CNG to value your collection for insurance purposes.

    Now for a few graphs…


    Interestingly, the ravages of Covid-19 don’t seem to have reached CNG auctions. Apart from one very rich auction in January, hammer prices have been on a slight upward trend. The same cannot be said for estimates – these are flat.


    This is illustrated when we look at hammer prices as a % of estimates. There is an upward trend, with hammer prices increasing more than estimates. But why is this happening?


    Just as there is an upward trend in hammer prices over estimates, there is an upward trend in the % of lots being sold. The correlation coefficient* (bear with me here) between hammer prices as a % of estimates and the % of lots sold is 0.73. Anything over 0.7 is usually considered strong. This means more lots are sold the lower the estimates turn out to be.

    At this point I should say ‘correlation doesn't prove causation’. We don’t know that lower estimates cause there to be fewer unsold lots, they just happen to occur at the same time.**

    Now, not only do lower estimates seem to mean fewer lots go unsold, but even lower estimates could mean even fewer lots go unsold. That may seem obvious, but it means there is an incentive for CNG to lower their estimates even further. If so, where will it end? With hammer prices rising, and so commission increasing, there’s no reason to stop now.

    But that does make the ‘estimates’ redundant for buyers and, dare I say, somewhat misleading. Estimates have become nothing more than a marketing tool offering discount prices that don’t have to be honoured.

    Obviously, this is not only a CNG phenomenon (they just happen to kindly publish their results). But why do they persist with valuations at all when they’re so out of alignment with the results? Katz start all their lots at EUR 5 and provide no estimates – yet bid prices soon shoot up long before the live auction begins. Given the incentives and the fact that CNG’s ‘estimates’ are not estimates at all, perhaps that’s where they’re heading too.

    See you at auction 476…

    Source: CNG, ‘Prices Realized’

    *The correlation coefficient is a value between -1 and +1 that shows how strongly two things are related. A value of +1 is perfectly correlated i.e. an increase in one goes with an increase in the other. A value of -1 is perfectly correlated negatively, i.e. an increase in one goes with a decrease in the other. A value of 0 shows there is no relationship between them at all.

    **It’s also true to say that if you sell more lots your combined hammer price will go up. It may be that the increase in the % of lots sold is causing the gap between hammer prices and estimates to grow. But total estimates are getting lower while the % of lots sold increases, so that seems less likely (as well as being harder to explain).
    finny, GregH, fomovore and 8 others like this.
  2. Avatar

    Guest User Guest

    to hide this ad.
  3. pprp

    pprp Well-Known Member

    I recently received the catalog of the feature auction 115. It has a strong chemical smell and found its way straight to the bin like the rest of the 2 or 3 recent ones. The Greek section looks unimpressive, the coins look dull, it must be a combination of bad photography bad printing and lower -than usual- quality coins. Another nasty surprise is the extra 2.5% of fee for participating live. If they continue like this they will end up lower than their low estimates.
    John Conduitt likes this.
  4. medoraman

    medoraman Supporter! Supporter

    I just chalk it up to the fact auctions historically was the wholesale market. Not sure everyone remembers that. Its like cars, the auctions are wholesale and dealer shops are retail. Retailers go to auctions to get stock.

    The owner of CNG graciously came here on CT and informed us the estimates are based upon wholesale values. Makes sense to me, most dealers sidelined from shows, etc, and more collectors are going directly to the auctions.
    Ryro, benhur767, DonnaML and 5 others like this.
  5. Restitutor

    Restitutor Well-Known Member

    This post is everything I dream about, I commend you on your excellent graphs and analysis!

    I bought 3 coins in the recent auction all well above estimate. Since I can’t travel from COVID, all my travel budget has spilled into ancient coins. I imagine I am not the only one, and that’s probably a reason why we are seeing these huge price upticks. From there, I wonder if CNG is also pushing out their highest quality coins to make even more money! I would probably do that if I ran the house!
 and John Conduitt like this.
  6. Active Member

    Fewer unsold lots, generating more bids, attracting more bidders and getting them hooked in the hope that they will eventually bid more are only a few reasons.

    Another major reason is that the greater the difference between estimate and the hammer price, the more successful do those auctions look to potential consignors. Who doesn't want to consign their coins to an auction that sells them for double their "estimate"? It's all marketing.

    You shouldn't care too much about their estimates. Estimate the coins yourself.
    Carthago, Alegandron, Ryro and 7 others like this.
  7. dougsmit

    dougsmit Member Supporter

    All this would be unnecessary if they would publish minimum bids rather than estimates. Of course that would still run foul of those lots with hidden reserves higher than 60% of the estimate but it would be less confusing. I do wish they would publish a summary of lots that sold under estimate so we could feel bad we missed them.
    This may be the best advice on the subject. If you don't understand why one coin sells for twice another, you probably should spend more time studying and less time bidding. There are other sellers that regularly estimate coins at twice what anyone would pay and end up with a pile of unsolds that get wholeesaled to little dealers. I must have bought dozens of coins from the late Don Zauche at shows that were in recognizable flaps trimmed down to remove the name of the sale (which was easy enough to track using acsearch - a very useful tool). I prefer the CNG method.
    PeteB, Ryro, Valentinian and 4 others like this.
  8. Finn235

    Finn235 Well-Known Member

    I don't pay any mind to estimates. Honestly, I've seen so much volatility in ancient coin prices that I think estimates are meaningless.

    Personally, I like the auctions like some of the lower end ones on Biddr (Zeus, Ares, Themis et al) who start all of their coins at $1-5 and let the market sort things out.
  9. Pellinore

    Pellinore Supporter! Supporter

    Estimates are always on a low level. I'm not much motivated by estimates, but by what I want and what I can spend. When I want some coin very much, a low estimate might help me winning it.
    Alegandron and John Conduitt like this.
  10. robp

    robp Well-Known Member

    I'm not convinced by the graphs above as they cover too short a period. December to March is essentially pre-Covid, but since then prices have appreciated because the only place to buy coins for the collection (or stock if you are a dealer) is at auction. Everyone has to go to on-line auctions to buy, so Covid has produced a market boost. More buyers = higher prices, possibly due to an increased number of rose-tinted spectacles because people see what they want to see. A lot of people are off work with more time to occupy. There are many factors coming together here capable of producing better prices.

    On the other side of the coin you say that estimates are too low. This is a case of trying to find a number high enough to keep sellers happy, whilst at the same time providing bidders with a possible bargain which will encourage them to competitively bid prices up. If the estimates are too high, then you end up in the position Baldwin's found themselves a few years ago. Estimates were essentially set at retail with opening bids 20% below. The result of that was you put in a bid at opening level if you wanted the coin, as you would then pay retail after commission was added. If you really wanted the coin you might bid an increment higher and pay over 'book' (whatever that means given no two coins are equal). An additional effect of the higher estimates was a higher pass rate because it could no longer allow for overgraded coins, nor the psychologically defeated buyer with the smaller pot of money to play with who will more readily say 'I can't afford that'. A lower estimate will frequently produce a higher end price because it produces more prospective bidders, some of whom will get drawn in. An estimate is also likely to reflect the 'book' value to some degree, which essentially means flat estimates throughout the year, rebased following the yearly price update.

    Estimates rarely reflect what the market will pay on the day.

    For every collecting area there will be specialists willing to pay a [sometimes large] premium for a particular variety which may or may not feature in the main market price guides. For the specialist, a guide price is not as important, whereas the general collector will be looking for an acceptable example at something around the 'book' value - and will often lose out as a result.

    If only for the above reason it shows how important it is to know the market for the coins of interest. Once a collector knows the market they can and many will happily bid maximum values that are both lower than estimate, or several multiples of it. It's horses for courses.

    The point you make about insurance valuations and not wanting CNG to do it is a red herring. Sums insured will either be for purchase cost (probably index linked), or replacement cost (for which read market value) and would never be based on auction estimates.
    finny and John Conduitt like this.
  11. GregH

    GregH Well-Known Member

    One benefit (to CNG) of low estimates is they can tell their consignors that their coins hammered for x% in excess of their arbitrary estimate. Estimates are pretty meaningless to me.
    benhur767, Alegandron and Ryro like this.
  12. Roerbakmix

    Roerbakmix Well-Known Member

    As an epidemiologist, I really like these kind of data analyses. However, I believe there are some flaws in your statistics. First, since your aim is to infer causality and reject the null hypothesis (H0 = hammer prices increase and estimates increase accordingly; H1=i.e. hammer prices increase while estimates stay low), you have to correct for confounding. Next, since you're looking at the relation between this hypothesis over time, you have to think about time-varying confounders (e.g. auctions in winter, near Christmas tend to yield higher hammer prices than those in autumn). Some other confounders come to mind:
    • what is the distribution of coins in the auction? Some coins tend to be more popular than others for no particular reason (at least, not a reason I've found out). This is probably unmeasured confounding, but you should be able to correct for the proportion of e.g. greek coins, roman coins, etc.
    • More importantly: what is the distribution of estimates in the auction. E.g. an auction with mainly low estimates may have a different estimate:hammer-price ratio than those with mainly high estimates. In fact, I looked into the effect of estimate on hammer price in a recent CNG auction (copied from a post here
    Here we see the relation between the estimated price (x-axis) and the realized price (y-axis);(n=856 coins). The dotted 45-degree line would mean that the coin was sold at the estimated price (i.e. perfect agreement between estimate and hammer).
    So, eyeballing this graph, most coins are in the 0-5000 hammered price, and the relation between the estimated price and hammered price seems fairly linear. However, zoomed in at the 0-5000 price range...
    ... it becomes clear however that coins in this price range usually hammer higher than the estimate, with only a few hammering at or lower than the estimate. When looking at the relation between the estimate and the ratio hammer prize / estimated price, we see this again: most coins at the lower end result in a higher hammered price
    indeed, the ratio between the hammered prize and the estimated price in the 0-5000 estimate range is 1.67 (95% CI 1.62 - 1.71); for the 5000 to max estimate range, this is 1.24 (95% CI 1.16-1.32), an obvious difference (though the 5000 cut-off is of course arbitrarily chosen).

    Speaking about statistics, you've drawn a linear regression line through your data-points which is probably not a good fit. In the graphs above, I plotted a LOESS curve, which is a sort of modeled smoother based on the Y- and X-values in combination with a couple of other parameters, but other regression methods may be better suited for your data.

    Ideally, we would have a dataset with the following:
    • the date of the auction (should be easily extractable)
    • the estimated price
    • the hammered price
    • the number of active bidders (which CNG would probably not disclose) per coin
    • the type of coin (i.e. Greek / Roman republican / British, etc.) (less easily extractable, but doable)
    • the condition (should be extractable, but is determined by CNG)
    • the rarity (which is also difficult to determine)
    • provenance (but, which provenance yields higher hammer prices?)
    • etc. etc.
    Then, we would be able to conduct a multivariabel model, and try to infer a relation between the estimated price (corrected for all these confounders) and the hammered price. This obviously would be nice to do, but it's quite an amount of work.
    Carthago, Orfew, NewStyleKing and 2 others like this.
Draft saved Draft deleted

Share This Page