How does the “Whisky Pricing Machine” work?

As the data basis for the Whisky Pricing Machine, we use the secondary market price observations you find on Whiskystats. Note that we purposely exclude retail prices since they generally do not reflect real transactions but only asking prices.

The Whisky Pricing Machine consists of a time series and a regression component. We reflect the direct price observations of “very similar” whiskies with the time series component. The regression component is learning how secondary market prices depend on the major whisky features in general (e.g. the brand, bottler, vintage, age and many more). By combining these two components, we use direct price observations but “correct” them with more general price structures and vice-versa.