Machine Learning & Whiskey

The Secondary

What's it to you

There is a growing sense in the bourbon community that some outside force, be it advertising, flippers, or pure magic that is forcing the price of bourbon higher and higher. I figured that while I won't be able to directly study the effect of some of those things on the market, the least I could do is attempt to analyze it. In this post, I analyze the secondary market and the influence that time has had on bourbon's price and in order to get the ball rolling, I took a page out of one of my favorite blog posts of all time and make heavy use out of the goolge trends api: Why Can't Canada Win The Stanley Cup.

In order to study the secondary, I collected data from a few auction houses got rid of things that didn't make sense (dirty data, mislabeled months, etc) and piled it all together into a tidy dataset. The data collected from consists of 6749 bids on about ~4888 unique bourbons and ryes spreading across 12 years worth of history (2005-2017). A caveat on the uniqueness is that many of the bourbons and ryes are one offs or those made specifically for the japanese and european markets. The highest bid in the dataset was ~$8,779 for a 24 year old Bitter Truth Rye whiskey and the lowest for US export Sheffer Fort Bourbon at a whopping ~6 dollars.

So what does the bourbon secondary look like?

First things first, it's important to note that the big spirit companies have spent a lot on marketing over the past 10 years in an attempt to take marketshare from the beer categories (their number one market competitor). Here's Diageo, for example:

Thanks to Google Trends we're able to see the direct impact of that marketing since 2004 and quantify the interest in bourbon as measured by the relative number of Google searches. What this plot shows is that search interest is steadily increasing as more and more people are getting familiar with bourbon.

When we plot the normalized google searches alongside the normalized mean of each auction, what's clear is the interest in bourbon and the price of bourbon have finally converged. More people have gotten into bourbon and the cost to aquire the brown stuff has finally started to reflect the new, higher priced reality. Given the discrepency between price and searches during 2009-2013, one could say that with respect to interest, bourbon was a mispriced asset.

Bourbon's Future

Now that the basic analysis is out of the way, I want to take a look at all the final purchase prices for bourbons in a nice easy log transform. (Note, for the non-data people, log transformations are nice. Doing so makes it easier to view the data).

Fitting a fairly naive trend-line gives us a very interesting picture. It looks like the 'bourbon boom' as a whole is plateauing. It's almost as if we've reached equilibrium in the secondary market. (Or maybe no one really wants to pay thousands of dollars for arguably over-oaked juice). Perhaps enough people who would happily pay thousands for a whiskey have already tried those expensive whiskies and thought: 'you know what, once is enough'.

Extra Plots

For the sake of showing some interesting plots, here's Pappy, Four Roses, Hirsch, and Wild Turkey:

What's important to note is that as mad as some might get about higher end bourbons becoming more sought after, at least it's at the price level of Karuizawas or older Macallans. Here's a hot take: Pappy and Brora are priced similarly in auction markets - which, after having both - is pretty ridiculous (Brora is just better). So here it is. Buy Brora. It's priced well for it's hype.