I am torn between wanting to believe ( I guess in all of us is the hope that something will work) and the proof I have actually seen.
A member of MT some time ago....cant remember who.....bought 10 yrs of tick data at some crazy cost from a bank to get pure (as pure as you can) data as that is what is important.....
He then tested something like 50 systems that had been programmed on 1 - 2 yrs data and were giving the developers the belief that they had finally found something that worked.....not over the top systems but just 20-50 pips per week systems.
The results showed that not 1 !!!! of these systems worked over the 10 yrs and that in something like 95% of the systems they were only profitable over the yrs they had been programmed for....i.e if the developer had 12 months data it worked on that 12 months.
why....dont know....but I guess it has something to do with, optimizing, curve fitting and the fact that 1-2 yrs is not as long as we think in
FX and that the mkt has subtle and not so subtle changes that basically end up ruining what looked like a winning system