There are many different ways to measure “best.” You could measure by daily PnL, return on invested capital, return on total capital including infrastructure and labor (programmers, traders, etc), or in other ways. Having said that, common automated forex strategies are pure arbitrage and market making. To write a successful automated forex trade using those strategies, you’ll want some extremely fast hardware, a bunch of extremely smart traders and programmers, and really great network connectivity to everywhere you want to trade. Think on the scale of your own dedicated line to UBS, not a “high speed” home internet connection to some forex site. In terms of startup costs, plan to raise a minimum of a few tens of millions of dollars to start, so that you can fund your infrastructure build-out, pay your programmers and traders competitively, and survive some losses as you develop your trades.