Hollywood Stock Exchange and Machine Learning

24 Jan 2018

The Hollywood Stock Exchange (HSX) is a fun little prediction market game that I have played on and off (mostly off) for about 7 years. In that time, I have grown my portfolio from the 2 million they start you with to close to 90 million. Most of that gain happened within the last 2 years, when I got back into the game due to a posting on the SA forums.

The premise of the Hollywood Stock Exchange is fairly simple. You bet on the 4-week (or 12-week in the case of limited release films) box office earnings of films. Each film has a stock value that represents the earnings, in millions, that the film is expected to produce. If you believe the earnings will be higher, you buy the stock. If you feel the earnings will be lower, you can short it. After 4 (or 12) weeks of showings, the film delists at whatever the box office total is at that point.

Now, there are some extra components to the exchange meant to help smaller portfolios become larger. Starbonds are the one I mess around with the most. In fact, Starbonds have been responsible for most of my portfolio growth over the past 2 years because of their ease of use and predictability. However, HSX caps the number of any particular Starbond that you can own, and since there are only so many adjusts occurring each week, the percentage growth sharply drops once your portfolio reaches a certain size. It’s fair to say that I have reached that point. A great adjust week may net me around 4-5 million, or about 4.5 to 5.5 percent of my portfolio. Though nice, most of that income comes from a small number of large adjusts, and keeping track of dozens or even hundreds of Starbonds is time-consuming and horrendously unfun.

As a result, I want to make Moviestocks a larger component of my portfolio once again. At one point, this was true. My thesis at the time was that the market often overvalued low budget films, particularly those with known actors. By investing a small sum of money in a very large number of these films, I could eke out positive portfolio growth without letting any particular film wipe out the gains of the whole port.