It sounds like (1) you really can't afford downtime or interrupted service/access/uptime, and (2) you personlly won't be available on location to tend to the hardware needs should a problem crop up.
(1) and (2) really don't mix well, it is a disaster in the making.
If you really need the reliability and near-perfect uptime but you also cannot be more available than whatever remote login connection you can arrange then you should really be thinking about the solution to your needs from an entirely different angle.
It really sounds like you need to be thinking of a solution in terms of
co-location, cloud servers, or
a VPS.
That way you are not responsible for maintaining the hardware, nor are you responsible for ensuring its uptime. That is all built into your monthly contract.
Since you are already planning to manage the compute aspects remotely, it doesn't really matter where the VPS itself is physically located so pick a provider that has the kind of reputation that you are willing to pay for.
You have so much going on, so much to deal with, on the business side of the equation (the forex analysis, etc) that dealing with the hardware side is just going to become a huge time-suck for you and quite possibly (as you are rightfully envisioning) could come to be the Achilles heel of your whole operation.
I used a VPS for a while. The only reason I opted away from that solution was because I work out of my home, all my hardware is on-site, and I keep a full set of redundant hardware onsite. I run UPS as well as have dual-ISP (one cable, the other DSL) for internet access redundancy. But that is something you can only manage when you are literally living 24/7 onsite with your hardware as I do.
If I couldn't live in the same building as my hardware then I'd only settle for co-location or VPS services as a substitute. Managing my own hardware from a distance is just not conducive to a high-uptime critical environment.