QuantConnect Community

It used to be harder but now we have the algorithm framework. One of it's features is to take multiple strategies (alpha modules) and combine the signals (portfolio contruction module) with one ordering mechanism (execution module). There are a lot of little details to work out but once you figure out the structure it's fairly straightforward.

I recommend building your own modules rather than just importing the default ones they provide. By importing the default modules you really have hidden 90% of your code and it makes for a very short and simple algorithm but it's not flexible at all. You can look at the source code on github as a starting point to see how the modules interact with eachother.

LukeI I may have misunderstood your reply, but in my mind, the algorithm framework is not meant to be a solution to running multiple strategies, at least not in the sense that I understand it. If you have two different strategies with a slightly overlapping universe, you need to decide whether they are similar enough to fuse into one algorithm. If the answer to that is 'no', then you should absolutely be running them as separate instances on separate accounts.

TedVZ unless your algorithm consumes too much CPU/memory to faciliate a second instance, I don't see why you couldn't run multiple instances on one VM - as long as they connect to separate accounts and have separate working directories, it should not be an issue. I don't quite understand your question about "handle the cases especially if the server/algorithm goes down and needs restarting" - running one or ten instances on a server shouldn't change how you handle restarting your instances. Are you planning on using Linux or Windows VMs? That will change how you approach the automatic restarting of instances - although both will probably involve turning the instance into some form of service, as that tends to be the easiest way to ensure uptime and reliability.