When I am approached to do a project I always tell the client that I myself trade. So what I suggest to them is to make the code highly configurable even if there's no need to. For example, instead of telling me that you are trading SPY with 50/200 MA cross over on 5min chart, I tell them I will create a program which reads parameters from file where they can put a range of values. Just tell me what to parameterize.Trust me, I am concerned about my ideas being stolen, but what options do I have, really?
Many times same strategy will work on some subset of instruments but will lose money on another subset.
One of the benefits is clients can actually run multiple instances side by side with different parameters.
The problem with this is that it makes the coding more complex, but I think it's worth it over a long time as market conditions change.
Of course if the developer is proficient with backtesting tools, given the concept, he/she can find some optimal parameters and instruments.
But, imho, there are very few people who have a fully automated (no discretion at all) system that is so unique that no one had ever thought of.
More likely it's something basic that had been refined over long time to suit their personality, risk parameters, and maybe just a small piece of a larger portfolio of uncorrelated strategies. Nothing that can make the developer a fortune by itself.
