It’s been about a year now since I started messing around with quantitative finance. I’ve made a decent amount of money, but nothing to write home about yet. (Turns out you need a lot of money to make a lot of money.) I’ve slowly realized that investing is more about risk management than it is about making sick gains. I thought I’d post one of the more profitable market risk algorithms I made.
I’m a big fan of the Google Trends website. You can see what people are looking up and do significant amounts of market research for free. I’ve been using it for my Bitcoin strategy. The theory was that if someone wants to buy bitcoin, they’d need to look up how to buy it first. Running the data through a Granger Test proved that theory to be true in most cases. It was also true for other crypto-currencies such as Ethereum, and Litecoin.
Here is the price of $BTC (green) compared with the data from Google Trends (Grey).
Google Trends defaults large data queries to months at a time for large time frames, so you need to normalize the data yourself to get a higher resolution. Pulling the data live is much easier, but Google normalizes peak searches to 100 no matter what your search window is and its a pain to transform it.
I understand that crypto has been a hot investment for a while now, so most of the gains can be attributed to the wild success of Bitcoin and its contemporaries. This strategy isn't designed to maximize gains. Its designed to evaluate the risk. I'm open to a debate on whether or not a strategy like this is viable, but it seems to be working well so far.