Automated Valuation Model

Our automated valuation estimates are based on a direct comparison between historical and recent auction sales data for artworks by the same artist. Using percentile-based price distributions, we match the user’s original input to a corresponding position within past sales and compare it to the equivalent position in the present market. This percentile-matching approach ensures that the estimate reflects overall market movement rather than isolated outlier results. Additional filters and normalisation steps are applied to correct for anomalies such as concentrated sales events or distorted pricing, allowing for more representative, time-sensitive valuations.

Unlike emerging AI-driven models, which can make unrealistic assumptions about what constitutes value in an artwork, our approach is rooted in statistically grounded methods. We focus on actual market behaviour — using percentile-based comparisons, structured historical data, and robust filtering — to produce value estimates based on real pricing patterns.

Input Parameters

Users select the artist’s name and medium and input a value and associated date for a specific artwork.

Comparable Market Sampling

Two time specific arrays of auction results are constructed; one reflecting sales activity around the user’s input date and another representing current market value for the same artist.

Percentile Matching

Each dataset is analysed to establish price percentiles. The user-submitted value is mapped to the nearest percentile in the historical dataset. The corresponding percentile is then identified in the current dataset. This percentile-based approach reduces the influence of extreme results such as unusually high value sales.

Value Adjustment calculation

The change in value between the matched historical and current percentiles becomes the basis for the revaluation.

This differential is applied to the user’s original input to produce an updated estimate aligned with broader market trends.

Market Representation and Confidence

The reliability of any estimate depends on the volume and balance of available sales. In periods where activity is limited or dominated by atypical works, the AVM will either apply adjustment limits or suppress the estimate entirely. Where the AVM is calculated, it is important to note that downward shifts may reflect a lower quality of works entering the market rather than a true decline in the value of the artist’s percentile. This effect is more likely when the number of observations is small.

Data Collection

Data Sources

We source data from a wide range of reputable public auction houses across multiple geographies, including [e.g. Sotheby’s, Christie’s, Bonhams, and capital city auction houses]. Each entry reflects a real, public sale of a specific artwork, with historical records going back to 2007.

We do not use private sale data or speculative estimates — every data point is based on documented public results.

What We Collect

For each sale, we capture a consistent set of attributes, including artist name, sale date, auction house, lot number, medium, price (hammer + premium).

Cleaning & Standardization

Auction house naming conventions and descriptions can vary widely. To ensure comparability, we:

– Standardize artist names and resolve aliases and different spellings

– Distinguish between original works and editions

– Exclude entries without clear attribution

Handling minor errors

While we aim for 100% accuracy, occasional discrepancies may occur due to inconsistent or incomplete source data — such as vague lot descriptions or mislabelled catalogue entries. In addition, although we apply a high level of care and oversight, some human error is inevitable in a manually audited dataset of this scale.

Crucially, these instances represent a very small fraction of the overall dataset. Our methodology focuses on robust sampling and trend detection, which ensures that occasional discrepancies do not materially affect valuations or analytical outcomes.

Data Coverage

The vast majority of significant artworks — whether by market-leading artists or those gaining critical recognition — tend to pass through larger, established auction houses. Focusing on sales at these major auction centres not only captures a statistically significant share of the market by value but also allows us to reflect meaningful pricing trends without distortion from lower-value or inconsistently recorded sales elsewhere.

Why It Matters

Reliable analysis depends on clean, structured, and audited data. Unlike raw auction feeds or scraped datasets, our system processes and reviews each auction lot individually. This labour-intensive approach ensures that every data point is standardised, categorised, and assessed for accuracy and relevance — resulting in information that is truly fit for purpose. The outcome is more meaningful comparisons, clearer pricing patterns, and greater confidence in our valuations.

Hammer Prices and Currencies

Revaluart records the sold price of each auction lot in the currency reported by the auction house. Where applicable, we calculate and remove the auction house’s buyer’s premium in order to arrive at the hammer price.

For currency conversion, we apply Bank of England exchange rates as of the day of sale. If the sale date falls on a weekend or public holiday, the next available business day’s rate is used. All values are then expressed in GBP for consistency across the dataset.

Glossary

U2D – Unique Two Dimensional artworks

This category includes both works on paper and works on board. As the Automated Valuation Model (AVM) measures change through price matching, it encompasses both paintings and drawings by the artist.

U3D – Unique Three Dimensional artworks

This category includes sculpture and installation works. As the Automated Valuation Model (AVM) measures change through price matching, it accounts for a range of three-dimensional practices.

Editions – Prints & Multiples

Editions refer to artworks produced in limited, pre-determined quantities, typically signed and numbered by the artist. This category includes prints, photographs, sculptures, and other works created using a reproducible medium.

Photography

Photography refers to artworks created using photographic processes, typically issued in limited editions and signed and numbered by the artist.

©2025 – Art Market Research Developments Ltd