Watch Appraisal Process

The following section covers the technical details and what-if modeling for one of Barker’s product offerings - watch price warranty for consignment.

Product Implementation

Barker's valuation warranty product offers consignors a policy ensuring they receive at least the quoted valuation for a sold asset. This policy would be offered on up to a 90 day window.

Covered products that sell below Barker's valuation, will result in Barker paying the consignor the difference between the sale price and the provided valuation. This valuation warranty product will be offered at the time of consignment.

Appraisal Process

Barker's valuation for each asset is a combination of a machine learning price prediction, and controls based on the assets evaluation at auction or other relevant market heuristics. The combination of these methods decreases downside risk associated with overvaluing a watch.

Barker's machine learning model predicts a watch’s fair market value based upon the key features identified for appraising a watch. For watch valuation Barker uses a CART (or decision tree) based model with a down-scaled prediction output to reduce risk, while ensuring valuations are high enough to be attractive to consignors.

A watch valuation, for example, works as follows:

• Customer submitting a watch to be sold at auction automatically receives valuation based on the information they provided

• The valuation provided is the minimum of:

  • Barker's machine learning estimate

  • Low estimate provided by the auction house or market heuristic relevant to that watch