I friend of mine is trying to apply his EA to the whole currencies in the terminal's List.
I've told him that using the EA in such a way, even if well configured for each currency, is not likely to provide profit.
Could you tell me your opinion? Is universal trading system realistic?
An EA should work well on one currency pair - then according to the portfolio principle, running it on multiple currency pairs will average the yield and break down the risks. But if it works bad on any currency pair, then by running on multiple pairs you will average the loss, even if you break down risks. However, there are differences from one pair to another. Some have swift moves - like EURUSD, others are more ranging like EURGBP or EURCHF...
But thinking in terms of replicating a one-asset strategy to multiple assets may be bad, because it seems to be fundamentally wrong to retailish trade mono-asset. I think that best trading strategies involve multiple assets at a time - like pair trading for instance, however these can't be backtested on MT4, hopefully on MT5 . If you would have such a system that works well with 2 pairs, it might work better with more than one combination, thus it could be "universalized"... But again, I repeat, a system must work pretty good itself, first before attempting to apply it simultaneously on other assets.
An EA that can trade anything would be a great asset! Realistically, that doesn't happen often because people tailor their strategies to a single pair's usual movements. If you can develop an EA that can trade everything with the same (or approximately) rate of success, then by all means do so. Everything being equal, I'd much rather have 10 pairs trading 1/10th of my balance than 1 pair trading my full balance.
Is universal trading system realistic?
It partly depends whether you're including synthetics but, even without that, "universal" is pretty unlikely. However, on the other hand, I know a lot of people who regard excellent performance on only one symbol (with notably worse performance on all other symbols) as prima facie evidence of curve-fitting.
It seems to me...
If you have a strategy that is sound, it will probably need to be adjusted/tuned/optimised to be successful for a particular currency pair's behaviour.
Most currency pairs' behaviours vary, so an optimised strategy will probably not be successful for long without re-optimising.
It should be possible for a good strategy to be optimised for a variety of behaviours (i.e. currency pairs).
So... if you have an EA that can self-optimise, it may well be successful on a large number of currency pairs most of the time.
No I haven't achieved it yet, but that's the plan ;-)
Thanks for your help.