CryptoRover

Between Sep 2017 and Apr 2018, I worked on “CryptoRover” (aka “CoinParse”). I never launched this product.

I was trying to predict the price of cryptocurrencies, primarily using social media sentiment analysis. It was a SaaS app where you’d plug in the currencies of interest, and you’d get “prediction alerts”. I learned a bit of ML doing it, which was good, but I stopped working on it for ethical reasons. The potential users seemed too unsophisticated that “do your own research” caveat wouldn’t have resonably prevented them from yolo-ing it all and then blaming me.

Another problem was, while it seemed that I was able to make pretty good predictions during “peacetime”, once in a while an outside event like “China considering banning Bitcoin” would completely throw a wrench.

While I never launched the product, I talked about it on Reddit, and and that’s how I learned about the kind of users it would be attracting.

My smarter friends suggested that if there really was something there, then I should prop trade, instead building a tool for your average crypto joe.

I’m now convinced that it would’ve been just a waste of my time in the grand scheme of things anyway, as trading is not a domain I wish to gain a lot of knowledge in.

Perhaps it was doomed for other more fundamental reasons, like choosing the wrong tool for the job, considering the noise in the data.

Had I done it as a prop trade / hedge fund, and if it were even moderately successful, of course the money would’ve been nice, especially to fund other projects. But I now really do think that evaluating projects solely by potential for financial success is the wrong way. The real question is if you want to spend your life working on that idea. Of course the potential financial success needs to be considered. But there are a million ideas that can make a lot of money which are not as pointless as trading is, in my opinion.

I’m quite happy that I came to my senses pretty quick.