I have read several posts on this and other forums and still being confused...I am looking for the best solution to speed up backtesting and optimisation for forex medium freq systems. (nb: just backtesting, real trading is executed on a different platform)The solution needs to handle tick data for at least last 5 years and allow for fast optimisation.I am considering Matlab, Mathematica, Tradelink or writing the code in Perl / Python / C++The objectives are:1. speed up the backtesting and optimisation2. achieve fast development of new strategiesAny suggestions? Thanks.

I personally think the code / system which you use to backtest should be the same exact code / system you use to trade your strategy except with a flag which determines if it's live or simulation. If you follow this approach then you know you will be trading what you are testing. If you use two completely different systems then the odds are you will have bugs and your model will not be executed correctly.So keep this in mind when making your decision. Matlab and C++ is what I use.