I originally wrote this as an answer to a question on Quora.
The world is littered with the empty wallets of engineers that approached the stock market as a math, algorithmic, or engineering problem.
The largest investment banks all have algorithmic trading programs, and they often lose large chunks of money. And sometimes they make large chunks of money. You might be able to compete in the gaps, but they have what amounts to infinitely deep pockets to hire the absolute best people and give them the highest-quality tools and set up servers that have sub-millisecond market access to systems you couldn’t begin to afford the connection fees for.
Something that makes money by trading algorithmically is one of the worst forms of parasite. Like a tapeworm, it removes nutrients from the market and offers nothing in return. But creating something like that is an interesting intellectual problem.
You might be able to build profitable trading software, but if you make it your life’s work, it’s entirely possible that you’ll end up with nothing to show at the end of the journey.