From my experience, I've never had a VPS provider that has 99.9999% uptime. I'd be thrilled if you could refer me to any provider that actually meets this. Thanks!
You're asking for a computer suggestion without giving any details whatsoever on the resources required to run your algorithms, which leads me to believe you aren't actually going to go through with this due to lack of experience. I'm going to recommend a raspberry pi so as to save you some money.
You mentioned 99.9999% on your post. Your company only claims 99.99% on its website (which I'm skeptical of). Once you get to this level of uptime, moving the needle a small amount is very costly. We pay a few thousand per month to AWS and even they only have an SLA of 99.95% on their VPSes (EC2).
That's horrible, you know that translates into 4.38 hours of down time a year right? If you're actually getting anywhere close to 99.95% you're getting a really bad deal; I use Liquidweb and haven't had a second of down time in nearly 3 years per my independent monitoring. A couple of major blowups like yesterday excepted (you could have worked around yesterday unless you had graphics stored on S3), I know AWS is getting far better than 99.95% in practice.
I am aware that 99.95% uptime allows for 4.38hrs/yr of down time. That's an SLA, not actual. SLA is just what they guarantee or they credit you for outages past the SLA's allowance. Also my systems run across multiple servers across multiple availability zones, so my actual system level uptime is greater than the uptime of any individual component. In any case, I am genuinely interested in VPS providers who can actually achieve 99.9999% availability at the VPS level. I didn't say I was only getting 99.95%. I said their SLA is 99.95% and liquidips's site says "datacenters with 99.99% uptime", despite his earlier claim that all VPS providers have 99.9999% uptime. In this case it's not saying SLA of 99.99%, it's saying "with 99.99% uptime", which is a more definitive statement. Far more than S3 was impacted in us-east-1 yesterday. I had alerts from AWS on 46 (S3 being 1 of the 46) different services being impacted although S3 got all the attention. Thankfully our end user facing systems aren't on us-east-1 (although we do have a jira and bitbucket instance hosted in us-east-1 that only suffered minor performance issues due to the negative impact on EBS). I think we understand each other. I was focusing on the claim that all VPS providers have 99.9999% uptime, despite his own company saying they have 99.99%. Also SLA is not the same as actual. I've had providers that have a SLA of 100% -- that just means they credit for any down time, not just that which exceeds 4.38hrs/yr for example. SLA of 100% certainly doesn't mean they actually provide 100% uptime -- no one can do that.
Great explanation, I think most aren't aware of those distinctions but you did a good job of explaining them. Back to the OP, I think in almost all cases you'll get better up time from a vps then a server at your house given that you've got your internet service to worry about, your power, and random stuff like your cat eating the cord or your kids spilling juice on the thing, but you might get lucky!
Take a look for Hydra app http://stocksharp.com/products/hydra/ Only as a home data feed server for live and history. Have some issues for scalability but for small team is best. Freeware.
renting a VPS close to NYSE/CME servers is one thing, but how do you compare/evaluate the latency from the VPS to the broker servers then to the exchange servers?