An auction estimate is built from two inputs: comparable sales of similar works, and the judgment of in-house specialists. The house then negotiates the published low and high figures, along with a confidential reserve, with the seller. So the number you see in the catalog reads closer to a marketing band than a neutral forecast, and the data shows it lands within range far less often than most people assume. In the first half of 2025, only about 10% of paintings at major houses sold inside their estimate range. The rest split wide. 62% beat the high estimate and 28% fell short, according to aggregated sales data compiled by Maddox Gallery. We think that gap between the printed estimate and the eventual price is worth reading carefully. It is a signal about how the house is pricing risk, and once you learn to read it, it tells you something the catalog will not.

How does an auction house build an estimate?

The starting point is comparable sales. A specialist pulls recent auction results for the same artist, ideally the same period, medium, and size, then widens out to similar works by quality tier. Sotheby's describes its own process as evaluating a work against "comparable sales and past performance of works by the same and similar artists" before issuing a non-binding estimate to the seller.

From there, the specialist adjusts for object-specific factors that comparables cannot fully capture: provenance, exhibition and publication history, condition, rarity, and where the work sits in an artist's output. A fresh-to-market painting from a sought-after series prices differently from a tired work that has cycled through three sale rooms in a decade, even at identical dimensions.

This is roughly the discipline we use when we underwrite a purchase, so it is familiar territory. We think about an acquisition as a two-step process. Step one is getting the artist market right, because even the best example by an unknown artist will probably wind up not being worth much. Step two is sourcing the best example at the lowest price. The same logic separates an A example from a C example: a strong canvas in a sought-after series and a minor work by the same hand can carry similar dimensions and very different prospects. Specialist judgment is what closes that gap. For important works, houses consult multiple experts and, in some categories, global teams. The specialist is also modeling bidder behavior: how many serious buyers exist at each price level, and whether a single determined collector or institution is likely to chase the lot. None of that lives in a comparables table.

Is an auction estimate a forecast or a marketing tool?

Once a house has an internal view of fair value, it decides how to position the public estimate, and that public estimate is a strategic choice. Artsy's analysis of the process puts it bluntly: estimates are "primarily marketing tools," only loosely tethered to fair market value.

The mechanics reward this. The low estimate works as the anchor. Christie's notes that bidding "generally opens below the low estimate" and climbs in increments of up to 10%, which makes that figure the psychological entry point for the room. Set it low and you pull in more registered bidders and more competition. Set it high and you risk a quiet room and an unsold lot. So specialists tend to push estimates down toward the floor of their internal range, because a conservative number creates the sense of a bargain and a better post-sale story when the work clears its high estimate.

Sellers usually pull the other way. Many demand the highest possible estimate on the theory that a bigger number means a bigger sale. Artsy calls this a paradoxical interest, because inflated estimates often suppress bidding and raise the odds of a buy-in. There is a third force. When the major houses compete for the same consignment, they sometimes inflate estimates to win the mandate, knowing sellers gravitate to the highest headline figure. The published estimate is the settlement of that tug-of-war.

Sitting underneath the estimate is the reserve, the confidential minimum at which the house may sell. Sotheby's sets its reserve at or below the low estimate and keeps it private. Artsy reports that in most cases the reserve falls between 50% and 100% of the low estimate. If bidding stalls below the reserve, the lot is bought in, meaning it goes unsold. So when a work hammers right at its low estimate, it likely cleared the seller's floor by a hair.

Phillips made the low estimate's role as an anchor explicit in 2025. Its Priority Bidding model ties a fee reduction to written bids placed early at or above the published low estimate, as reported by Observer. The estimate methodology did not change. The house simply built a financial incentive around the number bidders already treat as the reference point.

How often do auction estimates actually miss?

If estimates were neutral forecasts, most lots would clear somewhere inside the printed range. They do not. The 2025 data shows a market that mostly resolves outside the band, in both directions.

A Spring 2026 Merrill report on global fine art auctions at the three major houses found that 53% of lots sold above their presale estimates in 2025, up from 48% in 2024. The painting-only picture is sharper. Maddox Gallery's Q4 2025 analysis of aggregated sales for the first half of 2025 found 62% of paintings sold above estimate, 28% sold below, and just 10% landed within range.

Read that distribution carefully. A 10% in-range hit rate does not make estimates useless. Statistically, the estimate remains the single strongest predictor of hammer price. One machine-learning study built on roughly a decade of auction records found that house estimates alone explained about 94% of the variation in log hammer prices [NEEDS UPDATED DATA: figure from a 2019 Stanford CS230 study; confirm against more recent modeling]. The point is narrower and more useful. Houses now price most works deliberately below where they expect them to sell, so the printed estimate functions as a floor to bid against rather than a midpoint to land on.

That skew is a product of the cycle. Fine art auction sales fell to a decade-low average price of $25,329 in 2024 before recovering to $26,686 in 2025, a 5.4% gain, per Artnet. Total fine art auction sales reached $11.7 billion in 2025, up 13.3% off that depressed base. The three major houses booked a combined $4.55 billion, up 11.1% and their first growth year since 2022, according to Merrill. We have watched this pattern before. Coming out of a soft market, houses lean on conservative estimates to coax sellers back and rebuild bidder confidence, which mechanically pushes more lots above their printed highs.

Why do the estimate and the hammer price drift apart?

Several forces open the gap between what the catalog says and what the gavel records.

The first is strategic under-pricing, described above. In a cautious market, a low estimate is the cheapest way to manufacture competition. When it works, the lot sails past its high estimate and the house gets a "sold above estimate" headline. When it fails, the low estimate was still too high for the demand in the room.

The second is anchoring. Bidders lean hard on the low estimate as a reference point, and hammer prices cluster just above it. Research on estimate bias finds that raising an estimate does not lift the final price proportionally and can even depress sell-through, which is the opposite of what a seller chasing a high number expects. The estimate shapes demand as much as it predicts it.

The third is selection. Academic work on whether art auction estimates are biased finds that once you account for unsold lots, estimates look upward biased for the full set of consigned works. Sold lots tend to clear above the low estimate, which makes estimates look conservative and accurate. But the works that fail tend to carry estimates set at or above where any buyer was willing to go. You only see the prices for the works that sold, so the published track record flatters the house.

The fourth is concentration at the top. High-end demand stayed thin into 2025. Public transactions above $10 million fell 44% year over year in the first half of 2025, and there were zero lots above $50 million versus 13 in the first half of 2022, per Bank of America. The trophy market is where estimates miss most violently. In May 2025, a Giacometti sculpture carrying a roughly $70 million estimate went unsold and a Warhol estimated near $30 million was withdrawn before the sale. At that altitude, a single absent bidder turns a nine-figure estimate into a buy-in. To be clear, this is the same wealth-at-the-very-top demand that drives the whole high end of the market. When the top 0.01% steps back for a quarter, the trophy lots are the first to feel it.

Houses now hedge that risk with guarantees. In the first half of 2025, 45.5% of post-war and contemporary evening sale lots at the three major houses carried a guarantee, up 13 points year over year, and guarantees covered 73% of evening sale value, per Bank of America. A guaranteed lot has a backer committed to buy at an agreed level, which lets the house publish a tighter, more conservative estimate. So the headline estimate you read on a major evening lot increasingly reflects a pre-arranged financial floor as much as an open-market guess.

How should an investor read an auction estimate?

Treat the printed range as a negotiated band rather than a price target. A low estimate signals the seller's likely floor and the house's invitation to start bidding. A wide spread between low and high, often 1.5 to 2 times, leaves room for a "beat expectations" narrative. An aggressive estimate relative to recent comparables can flag a competitive consignment the house may have over-promised to win, which raises buy-in risk.

Watch the patterns more than any single number. A run of similar works in one sale tends to see later lots clear weaker than earlier ones, a well-documented effect, so lot order matters. A work hammering at its low estimate cleared by a whisker. A work that beats its high estimate by a wide margin tells you the house priced cautiously, and not that the market suddenly revalued the artist. A buy-in at the top end says more about a thin bidder pool that night than about the work's long-term value.

For platforms and funds that buy at this level, estimate discipline is part of underwriting. We think about it in terms of price relative to fair value: if our team values a work at a million and we can buy it for 700,000, that is a good deal, and if it is asking 2 million, it is not. The printed estimate is one input into that judgment, never the conclusion. [NEEDS INTERNAL DATA: Masterworks' auction database, with 50M-plus records, could quantify how often acquisitions cleared within, above, or below estimate, and how price-to-estimate ratios for target artists have moved across the 2024 to 2025 cycle.]

The Bottom Line

  • Auction estimates combine comparable sales with specialist judgment, then get set as a marketing band positioned just above a confidential reserve, which sits at or below the low estimate.
  • The estimate is the strongest single predictor of hammer price, but houses deliberately price below expected value, so most lots resolve outside the printed range.
  • In the first half of 2025, only about 10% of paintings at major houses sold within estimate, while 62% beat the high and 28% fell short, per Maddox Gallery.
  • The share of lots selling above estimate rose to 53% in 2025 from 48% in 2024, a sign of conservative pricing in a recovering market, per Merrill.
  • Estimates miss most at the top end, where thin demand produced high-profile buy-ins in 2025 and pushed guarantees to 73% of evening sale value.
  • For investors, the gap between estimate and price is a signal about the house's pricing strategy and the depth of demand, and not a forecasting error to ignore.

Sources

  1. Sotheby's. "How Are Estimates Determined?" Sotheby's Help Center. https://help.sothebys.com/en/support/solutions/articles/44002297435-how-are-estimates-determined-
  2. Christie's. "Financial Information: Buying Guide." Christie's. https://www.christies.com/en/help/buying-guide-important-information/financial-information
  3. Artsy Editorial. "How Auction House Estimates Are Set." Artsy. https://www.artsy.net/article/artsy-editorial-auction-house-estimates
  4. Maddox Gallery. "State of the Art Market Today: Q4 2025 Art Market Analysis." Maddox Gallery, Q4 2025. https://maddoxgallery.com/news/472-state-of-the-art-market-today-q4-2025-art-market-analysis/
  5. Merrill. "Art Market Spring Update." Merrill / Bank of America, Spring 2026. https://www.ml.com/articles/art-market-spring-update.html
  6. Bank of America Private Bank. "Art Market Fall Update." Bank of America, Fall 2025. https://www.privatebank.bankofamerica.com/articles/art-market-fall-update.html
  7. Artnet News. "Fine Art Auction Sales in 2025." Artnet, 2025. https://news.artnet.com/market/fine-art-auction-sales-2025-2758603
  8. Observer. "How Phillips' Priority Bidding Changes the Auction Fee Structure." Observer, July 2025. https://observer.com/2025/07/art-phillips-priority-bidding-auctions-fee-structure/
  9. Art Basel and UBS. "The Art Basel and UBS Survey of Global Collecting in 2025." Art Basel / UBS, 2025. https://theartmarket.artbasel.com/download/The-Art-Basel-and-UBS-Survey-of-Global-Collecting-in-2025.pdf