Forget HFT by yourself, at least until you've done some things and learned what it really requires. The barriers to entry are multiple and substantial, and you

Forget HFT by yourself, at least until you've done some things and learned what it really requires. The barriers to entry are multiple and substantial, and you can't figure out how to eliminate one unless you're really good with the others. Like any substantial / difficult / well-rewarded pursuit, plan on this taking 12+ months to be even minimally viable. Minimum viable product = Interactive Brokers account, automated via one of their API wrappers, with custom logic. This will get you to where you can modify orders ~ 10x per second, which is often enough that you're going to learn so much you don't even know what the questions are yet.... If you do all that and want to press forward, usually the next correct step (again assuming you're writing ALL the logic, risk handling, etc.) is something like Lime. "backtest my algorithms"... historical data isn't free, nor is any decent source of it "execute trades quickly"... what's the definition of quickly? there are systems currently active that can process an execution and have a response on the wire in less than 25us... that's 25 millionths of a second for wire-to-wire processing... Realize this: you are expressing a desire to play one of the most difficult games on the planet. In this game you do not get to choose your opponents. However, you can know that it's very, very unlikely that you know or can do anything they don't or can't.