There is no continuous market here - only comparable sales, an appraiser's estimate, and an auction reserve that can fail to be met. The gap between what an item is 'worth' and what it actually sells for can be wide, and it is measured only at the moment of an actual transaction.
12 min read · Level 4
I think the single most misleading number in this entire subject is the pre-sale estimate, not because specialists are careless in producing it, but because its precision is entirely borrowed from a small, imperfect set of comparable sales dressed up in a confident-looking range. A stock's price is the honest output of thousands of people risking real money against each other continuously; an auction estimate is a handful of specialists' best professional judgement, updated only when a new comparable happens to come to market. Both are useful. Neither should be trusted the same amount. The investors who do best in this subject treat every stated number - an estimate, an appraisal, a platform's mark-to-market figure - as the opening line of a negotiation with reality, not the answer, and they specifically go looking for the failed sales and the quiet private-treaty deals that never make it into a headline auction report, because that is where the honest price usually lives.
Pricing in this subject has no central mechanism at all comparable to a stock exchange's continuous order book. What exists instead is a small set of imperfect proxies, each of which only produces a real number at the moment an actual transaction happens.
The most common proxy is the comparable-sale: finding recent, verified sale prices for similar items and using them to infer what a specific item should be worth. Three recent auction results for broadly similar works by the same artist, in similar condition and size:
Comparable sale 1 62,000 Comparable sale 2 58,000 Comparable sale 3 51,000 --------------------------------------------------------------------- Average 57,000
An appraisal built from these comparables might land at 57,000, but a genuine appraisal typically expresses this as a range - say 45,000 to 65,000 - rather than a single figure, precisely because no two objects, and no two buyers on any given day, are identical.
At auction specifically, the estimate range translates into two further numbers that matter enormously and are frequently confused with each other. The reserve-price is the confidential minimum the seller has agreed to accept, commonly set at or below the low end of the estimate range - if bidding does not reach it, the item is unsold, or "bought in," regardless of how many people were bidding. The hammer-price is the winning bid at which the auctioneer's hammer actually falls - the number everyone in the room hears announced, and the number reported in sale results, though as lesson five shows, it is not the number either the buyer actually pays or the seller actually receives once fees are included.
Outside the auction room, a private-treaty-sale - a negotiated sale directly between a seller (often through a dealer) and a specific buyer, with no public bidding - is common for higher-value or more sensitive items, precisely because it avoids the public record and the risk of an item failing to meet its reserve in front of an audience. Private-treaty prices are rarely disclosed, which makes the comparable-sale data available for future pricing thinner than it already is.
A persistent bias runs through nearly all of these proxies, worth naming directly: appraisal-smoothing. Appraised and platform-stated values tend to lag reality and to understate genuine declines, because an appraiser working from infrequent comparable sales in a thin market has every incentive, conscious or not, to anchor near a prior valuation rather than move sharply on the strength of one or two recent, possibly unrepresentative data points. The result is that valuations in this subject tend to look artificially stable and mildly upward-biased compared with what a real, urgent sale would actually achieve on a given day.
And underneath every one of these numbers sits a liquidity-premium cost that a listed security's price does not carry: because it can take months to find the right buyer, a seller who needs cash quickly typically accepts a real discount to the estimated value simply for speed, which is a cost of illiquidity that never shows up in a stated appraisal at all.
The comparable-sale method matters as the least-bad option available, but it depends entirely on how genuinely comparable the comparables are - and in a market this thin, "comparable" often means "the closest thing anyone can find," which for a truly unique object like a specific painting or a one-of-one collectible can mean stretching the definition considerably. A share's price is set by thousands of participants trading an identical, fungible instrument continuously; a painting's estimate is set by a handful of specialists reasoning from a handful of imperfect analogues.
The reserve price matters because it is the single most consequential number in an auction that the public never sees in advance, and a failed sale - an item that does not meet its reserve - is itself informative in a way that rarely gets reported with the same prominence as a successful one. An item that fails to sell at auction has, in effect, just received a real, negative price signal: genuine buyers in the room were unwilling to pay even the seller's minimum, which is a stronger piece of evidence about current demand than any appraisal produced beforehand.
Appraisal-smoothing matters because it quietly distorts every historical performance figure this subject's marketing tends to rely on. An index built from appraised or platform-reported values will show lower volatility and steadier apparent gains than the same assets' actual realised sale prices would show, for the simple reason that appraisals move less and later than real transactions do. This is precisely the effect responsible for the low correlation with financial markets that alternative-asset marketing often highlights - some of that low correlation is genuine, and some of it is an artefact of a valuation process that simply does not update as often or as sharply as a continuously traded market does.
And the liquidity premium matters because it means the "fair value" quoted by an appraiser is, almost by construction, a value available only to a patient seller willing to wait for the right buyer at the right moment - not a price available on demand, which is exactly the assumption an owner needs to challenge before treating an alternative asset's stated value as money they could access on short notice.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
One painting, tracked through an actual attempted sale, showing how the estimate, the reserve, and the eventual outcome can diverge.
The estimate. Based on three comparable sales of 62,000, 58,000 and 51,000 (average 57,000), the auction house sets a pre-sale estimate of 45,000 to 65,000.
The reserve. The seller and the auction house agree a confidential reserve of 42,000 - below the low estimate, which is a common practice to protect against a genuinely disappointing sale day while still giving the item a real chance to sell.
Sale day, attempt one. Bidding opens at 30,000 and stalls at 38,000 - below the reserve. The auctioneer declares the lot "passed," and it does not sell. No hammer price is recorded publicly in most sale-result databases beyond the fact that it failed to sell, and the owner is left holding the same painting, now with a real, if quiet, market signal that current demand sits meaningfully below the pre-sale estimate.
Six months later, private treaty. Rather than risk a second public failure, the seller works through a dealer to negotiate a private-treaty sale directly with a specific interested collector. The agreed price, disclosed only to the two parties and their advisors, comes in at 40,000 - below even the failed auction's reserve, but achieved without the public record of a second passed lot, which the seller judged would have damaged the painting's saleability further.
The honest final number for this painting was not 57,000, and not even the 42,000 reserve - it was 40,000, achieved quietly, six months after the estimate was first produced, and nowhere near the number a casual observer of the auction-house estimate would have assumed the piece was worth.
Treating a pre-sale estimate range as a reliable prediction of the eventual sale price, rather than as one specialist's informed guess based on a small number of imperfect comparables.
Assuming a failed auction (an item passed for not meeting reserve) simply means the reserve was set too high, rather than treating it as genuine, negative information about current market demand for that specific item.
Using appraisal-based historical return series to judge an alternative asset's volatility or correlation with other markets, without accounting for the smoothing effect that infrequent, judgement-based valuations introduce compared with continuously traded prices.
Forgetting that a stated 'fair value' assumes a patient seller and an available buyer - an owner who needs to sell quickly should expect a real discount to that value simply for speed, which the appraisal itself will not mention.
Check how similar the comparables actually are in size, condition, date, and provenance - a wide estimate range is often a sign of thin or poorly matched comparable data, which is itself useful information.
A pattern of passed lots for comparable items is a stronger signal about real current demand than a fresh appraisal, and it is rarely advertised, so it has to be asked for specifically.
Estimate realistically how long a sale process would take and what a motivated, time-pressured seller would likely have to accept, rather than the headline value assuming unlimited patience and a perfectly matched buyer.
Ask whether a quoted historical performance series is built from appraised values or from actual realised transaction prices - the two can tell very different stories about how stable the asset class really is.
Before relying on any single appraised value, before bidding at auction, and whenever comparing a historical return series for this asset class against a continuously traded one.
Do not treat a single failed auction as proof an item is worthless - timing, presentation, and the specific bidders in the room on a given day all affect a single sale's outcome, and a genuinely valuable item can fail to meet reserve for reasons unrelated to its underlying quality.
It is a recurring pattern for a passed lot at auction to receive far less attention, in both public reporting and casual conversation among collectors, than a successful sale at a strong price. A single dramatic result for one work by an artist tends to be repeated and cited for years, while several quieter passed lots for comparable works by the same artist, in the same period, often go unmentioned entirely. Anyone relying on word-of-mouth or headline sale results alone is, without realising it, seeing a survivorship-biased sample of that market's real recent performance.
Ask specifically about recent failed or passed sales, not just successful ones, before forming a view of current demand for a specific type of item.
Create a free account to read the rest and apply it to your own investing.