I had an idea and I thought it could generate some interesting debate. Apologies if this has been done before.
First off, I'm not really interested in doing this, partly because I don't have the time, and partly because my trading style and development relies heavily on there not being anyone else involved.
But, this may throw up some interesting concepts.
So, imagine 20 good automated developers (from here ;-) got together to cooperate on developing an automated strategy. The particular methodology is not that important, but I imagine shorter term scalping would work better.
But the whole objective of the strategy is for the 20 people to each run it concurrently, and for the trading methodology to take advantage of this fact. I can imagine it would be very useful for a strategy to know what 19 other traders (each trading 5-10 lots?) are about to do, and to be able to rely on that.
At the simple level, moving the market could be a factor (is this legal?) but a clever strategy would be playing some of the games that are played in the order flow.
One advantage this cooperation has, even over the big banks, is the multi user nature. Even if a few weren't running, there are enough users to maybe be noticed pulling and placing liquidity. The big banks can certainly put volume on the bid/ask, but they can't easily be many users. Personally I don't know if there is something worth discussing there, but maybe there is.
What I'm talking about though is less trying to move the market by getting lots of users to do the same thing, and more the collaberative nature of this setup providing a niche that has not yet been exploited, and opportunities related to the structural concept that are not yet clear. Not to mention, what 20 good trading programmers could come up with pooling their experience.
As I said, I'm not really looking to do this, but I could easily imagine learning some new concepts or opportunities by a discussion of it.