I think that you can improve risk-adjusted performance greatly by accounting for correlations. RB, CL and HG are correlated with ES. EUR/USD is correlated with AUD/USD. CAD/USD is correlated with ES. However, correlations do change with macro conditions, so I'll be interested to see how you adjust.
Thank you very much drm7. You are right. At the moment i am proceding as follows. The bot computes "correlations" between all <b>pairs of instruments</b>. [ Computations are carried out every 15 minutes (adjustable by user, in case) and considering the last 2 days (also adjustable). ] So this is in any case <b>realtime correlation</b> (or "cointegration", if we prefer to call it that way: there is nothing linear in mkt!). I have defined, for now the following 2 folio rules (looking forward to ideas for new rules): [correlation rule] <b>1. if have a position >= K (packets) in any instrument "A", don't open a new position in any other instrument "B" with 0 position where the absolute correlation between A and B >= X</b> [optionally one can select positive correlations only, instead of absolute value] <i>(many thanks to Danny from Chicago for helping me formulating this rule! ;-))</i> [position rule] <b>2. if have a position >= K (packets) in any instrument "A", don't open a new position in any other instrument "B" with 0 position</b> [this second rule is more conservative and brutal, for people who really want to be prudential. Clearly, they might miss some hedging value of the folio. But again, is for when one is afraid about exposition] Let me know if you have more ideas for new folio management rules: i will add them. ;-) Tom
Wonderful work, I sent you the required email in your page. The philosophy behind it is rather intriguing. I have Tradestation, so I receive realtime data, do you believe it would be possible to use my data instead of IB? Thanks
Hi bodhidharmazen. Thank you! At the present, the program connects to IB only an relies stricly on <b>realtime data and events</b> received through the IB API. There are no plans to connect to any other broker or data source. (When something works perfectly, there is no need to mess with it ;-) Tim
Hello clarodina, sorry i have been away for a while, immersed in R&D and trading ;-) I will probably resume soon testing of new strategies. This is a .NET <a href="http://www.datatime.eu/public/gbot/">project</A>, connecting to IB (Gateway) through their free API. Tom
This journal looks very interesting. You say that you have traded the software on a real account with real money using a scalping strategy. Have you got any stats on your real trades? i.e. what types of profits are you showing from what starting balance. The Gbot website says that it only works with IB. IB has a minimum account balance of $10,000. i.e. Is this enough to start Gbot with or would it need more starting bal?
Hi Renton, i will resume soon with new test. I am just consolidating the latest ideas. Yes i am currently trading with real money. In fact i made this application first of all for myself. I am doing very well ;-)) However i don't like making claims, i just let anyone interested to be part of the adventure, and anyone can take his determinations based on capital, intelligence, skills and so on. It's just the minimum to expect that a trader can decide by himself about this. ;-)) In my personal opinion 10K is not enough to algorithmically trade with any tool, including my application. I would expect a minimum at least ten times larger (and that would not be generally sufficient to trade the futures, as i have being playing here, unless several folio constraints are enforced) With a relatively small capital, one could start trading a folio of ETFs, using a very moderate packet size (maybe increasing in time) and distributing the risk over multiple instruments. Possibly focusing on volatility, as the range is usually "limited" (tvix, uvxy, viix, vixy, vxx, xiv etc or VM, VIX) or instruments with prices near end range, and forcing directional entries. Maybe i will later show some examples of trading approaches which might be meaningful for smaller capital. In the meantime one can practice and when he is quite confident of his possibilities and skills (not for delusional reasons), finding the money will not be a problem ;-)) Tom
Tom - thanks for the reply. Would be great to see a journal that focusses on the smaller capital using the ETFs & indexes that you've suggested. Would love to follow that. For me, 10-15k would be an acceptable starting balance for an investment into a trading bot/platfrom that I am unfamiliar with - but probably no more..