The stock market and dancing are often based on the same thing: irrational exuberance. It stands to reason that by plotting the “danceability” of music against historical stock market data, one can predict stock market ups and downs based on the degree to which people are getting down.
Based on the 84-year-old theory that the length of skirts correlates inversely to the performance of the stock market, the Danceability Index is built on the principle that a preponderance of dance-friendly music in New York City could indicate irrational exuberance on the part of investors.
Even weirder, it works.
To create the index, quant/hacker Joe Rothermich, chief investment officer at Natural Selection Financial, determined the five most popular songs in New York City as determined by Last.fm and plotted that against The Echo Nest’s new Danceability attribute, which assigns songs a number between zero and one based on their dance-friendliness. (Disclosure: The Echo Nest publishes Evolver.fm.) He presented his creation, the Danceability Index, at Boston Music Hack Day on Sunday.
“We can mine that data, as you can in finance, to come up with a rule,” said Rothermich during his entertaining presentation (audio below). His method involves purchasing shares in the companies included in the Standard & Poor 500 index and holding them until the residents of the world’s financial center start listening to music that’s too danceable.
“If the Danceabilty Index is greater than 0.6, then we sell,” he added. “Otherwise we hold the S&P 500.”
The results are startling, for those used to tracking soberer economic indicators. As this below chart indicates, this Danceability Index rule led to far greater returns than would result were one simply to hold the Standard & Poor 500 without correlating buy/sell decisions to the dance-friendliness of New Yorkers’ favorite music:
But before you cash your 401K and dump it into Rothermich’s tongue-in-cheek fund, called Dance Dance Investing, LLC, consider the caveat he issued to Evolver.fm:
“This is a good example of (over) fitting parameters to past data for the purpose of fun, and I don’t really recommend investing with it.”
Sure, that’s what they all say.
10/19 Update: Here are the slides from Rothermich’s Danceability Index presentation (.PDF), and here’s the corresponding audio: