New Research with Claude · Query millions of auction results from your own AI toolsGet Started

Methodology

This page explains how ALT/FNDATA produces the realized prices, medians, records and price indices shown on our brand and market pages. The goal is simple: every figure should be reproducible, sourced from real transactions, and comparable across brands, categories and time.

In one line: our numbers are what items actually sold for at auction, normalized to US dollars, summarized with the median rather than the average, and tracked over time as an index set to 100 in 2005.

The dataset

ALT/FNDATA maintains 19.24 million realized auction results across 37 collectible categories, sourced from more than 100 auction houses and marketplaces worldwide and reaching back to the late 1980s. Of these, 11.5 million are sold lots carrying a confirmed price. The record is refreshed continuously as new sales close, and each brand page is stamped with the month it was last updated.

What "realized price" means

A realized price is the amount a lot actually sold for, including the buyer's premium where the house reports it. We use realized prices, not pre-sale estimates and not asking prices.

Currency normalization

Auction results arrive in many currencies. Every price is converted to US dollars using the exchange rate nearest to that lot's own sale date, so a 2004 sale is valued at 2004 rates and a 2024 sale at 2024 rates. We do not restate historical sales at today's exchange rate, which keeps each year's figures faithful to the market as it stood then.

How a brand is defined

Each lot is attributed to a brand from its cataloguing. Because houses spell and structure names inconsistently, we consolidate variants into a single canonical brand before any figure is calculated. For example, "Christian Dior" and "Dior" resolve to one brand, as do "Bulgari" and "Bvlgari". A brand page aggregates every category the brand appears in, so a maison that sells across watches, jewelry and handbags is measured across all of them, with a per-category breakdown shown on the page.

The statistics we report

FigureDefinition
Median realizedThe middle sold price. Half of sales fall below it and half above. This is our headline figure because it is not distorted by a handful of record lots.
Average realizedThe mean sold price. Useful alongside the median, but pulled upward by exceptional lots, so we never show it alone.
90th percentileThe threshold below which 90 percent of sales fall, a guide to the upper end of the market without being a single outlier.
Record resultThe single highest realized price on file for the brand, with the lot, house and date.
Verified sold lotsThe number of confirmed sold, priced lots behind the figures.

We publish a brand page only when it clears a minimum sample so the figures are statistically meaningful, and we suppress any single category on a page that falls below a smaller per-category threshold.

The ALT/FNDATA brand price index

The index tracks how a brand's typical realized price has moved over time. For each year we take the blended median realized price across all of the brand's categories, then express it relative to 2005, which is set to 100. An index of 180 means the median realized price is 80 percent higher than in 2005. The year-over-year figure compares the two most recent full years.

We use the median rather than the average so the index reflects the broad market rather than a few trophy lots, and we rebuild it from the underlying transactions on every update rather than chaining prior values, so a correction to the data flows straight through.

Coverage and limitations

Corrections

If you believe a figure is wrong, or you represent a brand or house and want to discuss the underlying data, contact us at altfndata.com/contact-us. Every published figure traces back to specific transactions, and we will investigate.

Source: ALT/FNDATA, 19.24M realized auction results across 37 collectible categories, normalized to USD. All brands · Market statistics