Power outages is an annoying thing which I addressed even before starting to run the system, by buying a big APC UPS (rack mounted with extra battery modules, can run essential systems in the house for a day+), I need a constant power anyway, so it's worth doing. The large units cost a couple thousand, and the lead-acid batteries need replacement every 5-7 years, though, they now have new models with Lithium batteries which should last much longer.. Also the "Awesome Shed" would probably be even more awesome with solar panels on the roof and some battery storage inside, just a thought
Is there a specific reason, why do you want to have trading server at home and not in the cloud? My primary setup is in the cloud, my secondary backup setup is at home, just in case.
I've gone through this before, but briefly: - Switching costs are high, eg getting IB Gateway to run on a remote headless server is a faff - For the sort of storage I need it turns out to be quite expensive to rent a machine - My system trades slowly, so the costs of outages are relatively low - I have a fetish for physical hardware Rob
I am a happy and proud user of pysystemtrade in production since a few weeks ago, while I develop my own system (just for fun/because I can). Thank you for all your work and valuable books, I am so looking forward to the next one! PS I posted a link to your post in the github discussions section.
My system has been running a bit hot lately (well at least when compared to Rob's) and I'm trying to see why that is. I have pretty much the same account as Rob, same risk target (well, I'm 24%) and IDM 2.5. Now things that are different, I have only 55 markets and 3 rule families (ewmac, breakout, carry). However, glancing at the forecasts, I don't see my forecasts being much more extreme than Robs. My instrument weights are obviously much higher since I have less instruments, but also FDMs are a lot smaller. Here's what I got right now: Code: Total risk (annualized percentage): 7.08 Sum of abs notional exposure as % capital: 210.42 Sum of abs annualized risk as % capital: 29.69 Net sum of annualized risk as % capital: -1.07 Looking at the computed positions, I don't think this is an issue with DO - raw computed positions are just high. I guess the question is, Rob how are you hitting your risk target with that few positions?
What makes you think so? I just recently migrated VPS and getting IB Gateway to run on it was a breeze. There's guides online for it as well (it's gotten even simpler since that was written). Curious how much storage you use for the trading portion.
Just glancing at your risk report, compared to mine, you're running slightly more risk but as you say with a lot more positions. Total 5.8 (lower than yours but not by much) Sum of abs(notional exposure % capital) 75.4 (much lower) Sum of abs(annualised risk % capital) 13.8 (about half) Net sum of annualised risk % capital 5.8 Basically you have a hedged book with lots of longs and shorts and a fair amount of leverage, balancing out to not much risk, whilst I have more of an outright bet on, albeit not a very big one. This might be a consequence of your instrument choice. To take an example, suppose someone invested just in equities and currently overall equity signals were weak; with some markets short and some long. Their book and risk would look very much like yours. Whereas, take someone with a spread of asset classes, but where only one or two asset classes had a strong signal. Their book would look like mine. It also looks like you might have fewer problems with discretization than me, eithier because you have fewer markets or more capital, or more markets with small tick size. Rob
Whever I've tried it in the past I've found it fairly painful. Of course things are always improving. Rob