The art market publishes more data than it did a decade ago, but most of it describes the past, and most of the present stays hidden. A handful of institutions release free annual reports built on proprietary numbers. A few museums and universities open up historical transaction archives. A small group of paid databases sell the rest. The roughly 35% of the market that runs through public auctions leaves a visible price trail. The other 65%, sold privately through dealers and galleries, does not. For an investor, that split is the whole story. You can benchmark an artist's auction record with real precision, and you can learn almost nothing about what a gallery charged a collector last week.

This matters because data is what turns a collectible into an asset class. You cannot underwrite an allocation you cannot measure. We learned that the hard way. When we started Masterworks, we did not actually know how much art appreciated, because there was no reliable index methodology and no API to pull the prices from. We hired interns, bought thousands of paper auction catalogs, and recorded individual events of paintings purchased and then sold. So we have some standing to say where the open data is good, where it is thin, and what it changes for anyone treating art as an investment. Below is who publishes what, what remains locked away, and how the growing pile of open data shifts the calculus.

Are the free art market reports based on open data?

No. The most widely cited art market numbers come from reports you can download at no cost, written on data the public never sees. The flagship is the Art Basel and UBS Global Art Market Report, researched by economist Clare McAndrew of Arts Economics. The 2026 edition, the tenth in the series, was released in March 2026 and covers calendar year 2025. It put global art sales at an estimated $59.6 billion, up 4% year on year, across about 41.5 million transactions. It breaks that total into a dealer sector worth $34.8 billion, public auction sales of $20.7 billion, and auction house private sales of just under $4.2 billion. The United States held 44% of sales by value, with the United Kingdom at 18% and China at 14%.

The report is free to download after registration. The data underneath it is not. McAndrew builds the estimates from dealer surveys, collector surveys, auction databases, and government trade records, then publishes the aggregates. Individual prices, the survey responses, and the dealer level detail stay with Arts Economics and its data partners. The public gets the ratios and the direction of travel, and nothing more granular without a separate license.

The same pattern repeats across the field. Artprice, owned by Artmarket.com, published its 32nd annual report, "The Art Market in 2025," on 10 March 2026, covering global fine art auctions and reporting a 6.5% rise in the number of works sold. Bank of America and the research firm ArtTactic released an inaugural United States Art Market Report in 2026, the first study built specifically around the American market. Deloitte's Art and Finance Report, Hiscox's Online Art Trade Report, and TEFAF's segment studies follow the same model: a free PDF on top of survey data and commercial databases that are not released. These reports are genuinely useful for sizing the market and tracking sentiment. None of them hands you the raw prices.

What art market data is actually open and free to use?

The closest thing to open data in art sits in museum and academic archives, and it looks backward. The Getty Research Institute maintains the Getty Provenance Index, a free, searchable database of historical art market records drawn from European auction catalogues, dealer stock books, and archival inventories spanning the 16th through 20th centuries. It was built for provenance research and market history, and economists use it to construct long run price series. You can query and export subsets, though there is no single bulk download and the interface assumes a historian, not a quant.

The Getty also digitized the business records of the Knoedler Gallery, the New York dealer that operated from the mid 19th century until 2011. Those stock books and ledgers record artists, mediums, purchase prices, and sale prices, and academic teams have turned them into structured datasets released under open licenses. Several universities maintain similar art price datasets, often hedonic or repeat sales series built from old auction catalogues and published as supplements to economics papers. The long run return studies that put fine art's historical appreciation in the low to mid single digits annually rest on exactly this kind of archival data. [NEEDS UPDATED DATA: a specific, dated long-run art-return figure from a study published in the last 12 months would strengthen this claim; the foundational academic series (e.g., Goetzmann and collaborators) predate the 12-month window.]

The value of these archives is real, and so is their limit. They tell you what Impressionist canvases traded for across decades. They tell you nothing about a contemporary artist who came to market in 2024. And here the limit cuts deeper than it looks, because the appreciation lives in the recent end of the market. The work we study most closely, post-1970 Contemporary, has historically appreciated faster than Impressionist or Old Master work, and that is precisely the segment the open archives cannot reach. So the open data is rich exactly where the money is cold, and quiet exactly where it is live.

How much of the art market is private and unpriced?

More than half of it by value never reports a price. Using the 2026 Art Basel and UBS figures, the dealer sector alone was 58% of global sales, and adding auction house private sales brings the privately transacted share to roughly 65%. Gallery and dealer prices, both primary market sales straight from the artist and most secondary market resales, are negotiated in private and disclosed only to the parties. High end dealers routinely write confidentiality into their sales precisely to keep realized prices and discounts off the record, which lets them discount selectively and protect the stated value of inventory.

Even the visible third of the market, public auctions, discloses less than it appears to. Auction houses publish results for sold lots and headline sell through rates, but there is no standardized public dataset of buy ins, the works that fail to sell. Price discovery skews toward success because failed lots leave no print. Guarantees and third party irrevocable bids, the arrangements that effectively pre sell major lots before the auction starts, are flagged in catalogues with small symbols but never quantified. The houses do not publish the guarantee amount, the fee split, or who the guarantor is. A record price set on a guaranteed lot does not mean the same thing as a record set by live competitive bidding, and from public data you cannot tell the two apart.

Then there is the gap between the number quoted and the cash that changes hands. The hammer price is the winning bid. The buyer pays a premium on top of it. Sotheby's raised its buyer's premium to 28% on hammer prices up to $2 million in New York, effective 13 February 2026, with lower rates on higher tiers. Christie's charges 27% up to $1.5 million as of September 2025. A work hammered at $1 million can cost the buyer roughly $1.26 million all in, while the consignor receives the hammer minus a seller's commission that is privately negotiated and never published. Two lots with identical reported prices can carry completely different economics for the people on each side. Price databases that standardize on the all in figure build a systematic wedge between the number researchers treat as the transaction price and the actual cash flows of the trade. We think a lot of the confusion in this market traces back to that wedge.

Where can investors buy structured art price data, and what does it cost?

If you want comparable, structured price histories rather than aggregate reports, you pay for them. Three providers dominate. Artnet's Price Database holds more than 30 million auction results from hundreds of houses and markets itself as the most complete archive of auction prices, used by appraisers and dealers for valuations. Artprice positions itself as the world leader in art market information and runs long standing artist and segment indices on top of its auction database. MutualArt offers auction results, artist analytics, and alerts on a freemium model. Professional access to these tools runs from the low hundreds to the low thousands of dollars or euros a year, and bulk programmatic access, the kind serious quantitative work needs, requires separate and more expensive licensing.

The data these companies sell is itself a compilation of public auction results, copyrighted and guarded as private property. So here is the quiet irony of art market transparency in 2025-2026. The prices were public when the gavel fell. The organized, searchable version of those prices is a paid product. No open data initiative releases a full machine readable global auction dataset at any scale, which is more or less why we built our own. Masterworks studies more than 50 million auction records to look at artist performance, price dispersion, and historical returns, and that scale is what makes systematic comparison across artists and price bands possible in the first place. [NEEDS INTERNAL DATA: specific findings or metrics Masterworks has published from its 50M+ record dataset that could be cited here directly.]

Is more data making art behave like a real asset class?

Slowly, yes, though it is not making art liquid. Greater data availability is shifting art buying from intuition toward evidence. Repeat sales indices, which track the same object across multiple sales, are the cleanest tool the market has, because they sidestep the problem that no two artworks are identical. It is the same idea Robert Shiller used to build the Case-Shiller home price index: rather than averaging a basket of different things, you follow one object across its transactions and isolate the pure price change. The Sotheby's Mei Moses indices, which Sotheby's bought in 2016, follow roughly 45,000 repeat sales across eight collecting categories at Sotheby's, Christie's, and Phillips, with several thousand turning over each year. For an investor, a repeat sales index strips out much of the noise that wrecks naive comparisons of one off auction results. When the same Warhol trades in 2018 and again in 2025, the change in price tells you something real.

Machine learning models now estimate likely price ranges from variables like artist, medium, size, date, sale venue, and prior sale history, then flag lots that look mispriced. In 2025-2026 the practical use is decision support, screening acquisitions and stress testing portfolios, rather than automated appraisal. The hard limit is the one that runs through this whole subject. A model trained only on auction records is blind to the 65% of the market that trades privately, so its estimates are probabilities, not verdicts. Blockchain and tokenization are adding tamper resistant provenance and title records and opening fractional ownership of high value works, which can cut due diligence friction and reduce some authenticity risk, though they do nothing to settle valuation or guarantee that a secondary market will be there when you want to sell.

The net effect is narrower than the headlines suggest. Better data lets you examine turnover, repeat sale performance, category volatility, and concentration risk far more systematically than before, and it exposes how segmented this market really is, with blue chip names like Picasso or Basquiat behaving nothing like speculative momentum names. It also keeps confirming that large parts of the market are private and relationship driven. We would put it this way. Open data is closing the gap between narrative and measurable fundamentals. It is not closing the gap between art and a stock you can sell in a single afternoon.

The Bottom Line

  • Roughly 35% of the global art market by value runs through public auctions that leave a visible price trail, while about 65% trades privately through dealers and galleries that disclose nothing, according to the 2026 Art Basel and UBS report.
  • The most cited reports, including Art Basel and UBS, Artprice, Deloitte, and the new Bank of America and ArtTactic United States report, are free to download but built on proprietary survey and database inputs the public never sees.
  • Genuinely open datasets exist mainly in museum and academic archives like the Getty Provenance Index and the digitized Knoedler ledgers, and they cover historical transactions rather than the live market.
  • Even within public auctions, buy in rates, guarantee amounts, third party bid terms, and seller commissions stay hidden, so a reported price can mask very different economics for buyer, seller, and guarantor.
  • Structured, comparable price data sits behind subscriptions at Artnet, Artprice, and MutualArt that cost hundreds to thousands of dollars a year, with bulk access licensed separately.
  • More data, repeat sales indices, and machine learning are pushing art toward evidence led investing and clearer relative pricing, but none of it makes art a liquid asset or removes the blind spot created by undisclosed private sales.

Sources

  1. Art Basel and UBS. "The Art Basel and UBS Global Art Market Report 2026." Art Basel, March 2026. https://www.artbasel.com/stories/the-art-basel-and-ubs-global-art-market-report-2026?lang=en
  2. McAndrew, Clare (Arts Economics). "The Art Basel and UBS Art Market Report 2026." UBS, March 2026. https://www.ubs.com/global/en/our-firm/art/art-market-research/global-art-market-report-2026.html
  3. Artmarket.com. "Artmarket.com publishes its 32nd Artprice Annual Report, The Art Market in 2025." PR Newswire, 10 March 2026. https://www.prnewswire.com/news-releases/artmarketcom-publishes-its-32nd-artprice-annual-report--the--art-market-in-2025--showing-12-growth-with-the-usa-strengthening-its-dominant-position-plus-gemini-deep-thinks-ai-audit-of-artprices-strategy-for-20262030-302708675.html
  4. Bank of America Private Bank. "2026 U.S. Art Market Report: Economic Outlook and Trends." Bank of America, May 2026. https://www.privatebank.bankofamerica.com/articles/us-art-market-report.html
  5. The Art Newspaper. "Sotheby's hikes buyer's premiums as auction houses test new fee structures." The Art Newspaper, 17 February 2026. https://www.theartnewspaper.com/2026/02/17/sothebys-adjusts-buyers-premiums-fee-structures-securitisation
  6. Antiques Trade Gazette. "Sotheby's raises buyer's premium to 28% at the lower tier." Antiques Trade Gazette, February 2026. https://www.antiquestradegazette.com/news/2026/sotheby-s-raises-buyer-s-premium-to-28-at-the-lower-tier
  7. Sotheby's. "Sotheby's Acquires the Mei Moses Art Indices." Sotheby's, 2016. https://www.sothebys.com/en/articles/sothebys-acquires-the-mei-moses-art-indices
  8. Getty Research Institute. "Getty Provenance Index." The Getty Research Institute. https://www.getty.edu/research/tools/provenance/
  9. Family Wealth Report. "US Remained Largest Art Market In 2025, Art Basel, UBS Report 2026." Family Wealth Report, 12 March 2026. https://www.familywealthreport.com/article.php/US-Remained-Largest-Art-Market-In-2025-%E2%80%93-Art-Basel,-UBS-Report-2026-?id=207156
  10. Luxury Tribune. "Global Art Market Report 2026: Global Sales Grew by 4% in 2025." Luxury Tribune, 12 March 2026. https://www.luxurytribune.com/en/global-art-market-report-2026-global-sales-grew-by-4-in-2025