Best backtesting Period

Hi all,

I was just wondering what is a good time period to backtest an automated system. 1 year , 6 months ...... ?
Is it true that market dynamics completely change over time and backtesting too far in the past of a system that works presently is just an overkill or even misleading.

i consider how many trades are in the backtest than how much data there is. as long as i can look at about 300 trades i am cool with that. if its a scalping system, you may only need 3 months of data to get 300 trades, on a trend system you may need 30 years to get that many. hope this helps.

Thanks Merlin. It removed some major doubts in my mind because my system fails considerably prior to 300 trades .

Backtesting is just a waste of time and effort. no two periods in forex will be the same . so you might backtest your system from an 2009 to today and your system will 'perform' poorly then you backtest from 2005 to 2010 and your system performs well...