I'm going to say that I think structures along that line is a much better way of structuring things anyway, as I'm currently refactoring to do it it provides your algo code some independence from the broker API. But then, my Java platform got blown out of the water by my upgrading to a new API(it was using a much earlier API).
So now I'm refactoring IB's API Demo code with some of my platform code to run as an API Babysitter, sitting between their API and my platform. In future, it should be much easier to survive an API upgrade.![]()
Exactly! The trading logic layer algorithm should be separated as much as possible from the broker layer API. The algo only should be concerned that the market API manager layer (whoever it is) is going to work.
algo
{
marketAPISession.executeMarket("dax","long", 1000)
}
The algo doesn't care who is behind "marketAPISession".
(it was using a much earlier API).