Two days and no response. Does not instill a person with confidence. I must assume that vendors like MetaTrader are stealing your code the moment you use it. If you happen to stumble upon a truly winning trading system, they will know it the moment you test it. What happens next? Do the brakes in your car fail mysteriously?
It would take me or any other competent system designer than 2 minutes to develop a great system based on curve fitted back tests. How would meta trader know that the system they where seeing wasn’t curve fitted ?
Commercial EA vendors quite routinely develop systems that appear to be great in backrest by using these kinds of tricks.
The first problem they are going to have is identifying if the system is a legitimate method, or due to curve fitting, and that’s not quite as straightforward to do as it sounds,
The second problem is that by random chance, about 35% of forex accounts are profitable over any 3 month period. That means that 35% of people with zero skill make money just through luck! and that statistic also applies to expert advisors.
Establishing if the system has a genuine edge is a phenomenal amount of work, and I’d be amazed if the developer of a charting platform had the resources to do it.
There’s no doubt that many services where developed in the hope of identifying profitable systems, but by the time system designers are anywhere near reaching the point of developing profitable models, their code, or their results are generally not in the public domain.
I think the point that you raise is legitimate, and it’s probably best to do everything possible to hide intellectual property from potential threat, but I doubt that meta trader are routinely scanning accounts. I would however be amazed if they didn’t take a very close look at the EAs submitted to their annual auto trading contest
Just to add, meta trader of course already has the technical capabilities inbuilt to steal code if they so wished. The FTP facility used to upload statements to a domain of your choice could very easily be used to upload other files. I doubt they’d do it as all it would take is one person to discover that was happening and their credibility would be destroyed.
If you where paranoid i guess you’d protect against it during testing by maybe throwing in a bunch of completely random trades, or by fading 50% of your signals. You could separate component elements of your system testing entries,exits and trade management in isolation, and therefore not using profit or loss as a metric to assess success of your model.
If you have specific money management rules you could mix things up a bit by using multiple brokers so no one broker gets to see the full picture
The obvious solution is test across a range of platforms and API’s so no one provider gets to see everything
You can hide code in DLL’s or web services which is the approach I’ve always tended to use, so programs are just calling external functions. It’s not foolproof, but it’s better than disclosing everything in one easy to access file.
You shouldn’t forget that the broker has a really nice edge already, and they don’t really need anything else, and quite frankly, I’d be amazed if there are many systems out there that have a larger statistical edge than the broker already has available to them.