Currently, I'm just papertrading on a look-forward basis. It will be a lot of work to figure out how to backtest, but I guess I will need to do the legwork.Great stuff, I'm still being lazy about itHave you tried backtesting it at all to see how it behaved historically ?
Fully automated futures trading
- Thread starter globalarbtrader
- Start date
no, no emails, but will check..
I'm thinking to try this (backtest\paper first) with my regular execution algo (i.e. without limit orders).. The weird part about my system is that in real-time trading I'm generating 2 signals: one based on the bid price and one on the ask, and using the "worst" of them for the trading decision.. So applied to mean-reversion I'll only make sell decision based on the "bid-based" signal and a buy decision only on the "ask-based" signal, and then the execution algo might improve on the price.. Don't know if it makes much sense really..
I'm not following you. I don't think you can improve on the original design of GAT's MR system. You set the limit order (buy or sell) based on the price that would produce a position 1 contract above or below the current optimal position.
yeah, that's probably the best, but it will require more changes in my system, so at first I'll just try it with my current execution algo, but the signal will be generated not using the last\mid price but the bid or ask price (whichever is 'worse').. (my system is event-based so I'm processing every bid\ask change anyway).. I.e. I'll sort of wait for the price to deviate from the equilibrium even more (by the bid-ask spread) before entering into the position, which will probably result in fewer trades.. Maybe it's not going to work, then I'll try it as described in the book..
Last edited:
Rob, I'm going to try my luck and ask about investing in a trading discussion.. I just finished Antti Ilmanen's latest book and naturally have read Smart Portfolios and related blogs posts. Consider brave investor starting from scratch with long-only ETF's (no shorting, no leverage, no advanced trading methods). In order to harvest all the diversification benefits with given constraints, would you diversify developed equities into countries as in the book or would it make more sense to diversify into styles as much is possible (beta + carry, defensive, momentum, value) as in the following example? In addition, would you equal weight the styles? Of course we can't fully benefit from style premias without going short but at least the following is not 100% beta and it's cheap and easy to implement. There is a lot more beta per region than the ideal 20% but the question is which is less worse approach, by countries or by styles.
In theory this would be better (minimal loss from not diversifiying across countries versus factor exposure return pickup) but I doubt the slight benefit would survive the application of costs.
Rob
Hello Rob, could you confirm if your Combined Breakout strategy(without trend and carry) or Strategy 9 are profitable on other currencies future besides Japanese Yen and Euro FX futures? These are only things which turned out to be profitable with those two strategies. I thought it was very weird to have these results.
Why is it weird? Out of 100 or so futures tested, not all will be profitable.
Rob
Not because the strategy is not profitable on other currency futures but it actually works on Euro FX. Didn't expect it to be more directional than the other currency pairs despite its huge daily traded volume.