Fully automated futures trading
- Thread starter globalarbtrader
- Start date
Hi GAT,
Would be really interesting to hear your thoughts on system design from a coding perspective. I'm familiar with your initial pysystemtrade but lost track after a large refactor, i'm assuming it hasn't changed too much. What I really like(d) about pysystemtrade was the simplicity of defining the strategy in yaml, this then becomes the main interface when in user is in research mode.
What would you do differently in your next version? Do you think you got the abstractions right? How could they be changed to suit more trading styles?
Thanks for all the education, i've learned a lot.
Would be really interesting to hear your thoughts on system design from a coding perspective. I'm familiar with your initial pysystemtrade but lost track after a large refactor, i'm assuming it hasn't changed too much. What I really like(d) about pysystemtrade was the simplicity of defining the strategy in yaml, this then becomes the main interface when in user is in research mode.
What would you do differently in your next version? Do you think you got the abstractions right? How could they be changed to suit more trading styles?
Thanks for all the education, i've learned a lot.
What is other 50%?
I think I see the core of pysystemtrade still being used to generate (on a daily basis) some long run return expectations based on daily data. However there will have to considerable additional layers on top of that, and I'd probably replace the p&l engine for example to use intraday data.
GAT
Good question, tradrjoe. I use a simplified version of GAT's system, which excludes carry, and trades 16 instruments, and I am essentially flat in 2019. GAT, do you know what elements of your system have outperformed? Is it one or two outlying instruments or rules that I may not have? I would have thought I'd be pretty highly correlated with you.
It would be a fair bit of work for me to work that out, but I will see if I get some time to do it, as it would be interesting.
GAT
I have to ask - why intraday data for the p&l engine? Would you use this to override your system in extreme circumstances? Or do you have some other use in mind?
Good question Napoleon@Dynamite. I also use a system without carry, but on a smaller number of instruments (usually about 8~10 open positions). My account value remained mostly flat during 2019, with an exception in early January and the recent two weeks. On these occasions did my account value go down. Overall my 2019 YTD return is thus negative. So I was surprised to see GAT be so enthusiastic. On previous occasions my system seemed to perform more or less in line with what GAT reported, but this time it seems really different.
https://on.ft.com/2QgrN2H
This is quite an interesting read. One hilarious quote from Harding below:
Trend-following is now delivering “a pretty uneconomic level of return”, says Mr Harding, who is now chief executive of hedge fund Winton Group. “Certainly not enough to justify being a big swinging dick hedge fund.”
This is quite an interesting read. One hilarious quote from Harding below:
Trend-following is now delivering “a pretty uneconomic level of return”, says Mr Harding, who is now chief executive of hedge fund Winton Group. “Certainly not enough to justify being a big swinging dick hedge fund.”