RichardL Veteran   Posts: 134 Joined: 9/18/2004  User Profile | as I understand it: (i) more up to date data, instead of the Back Test being between 1-2 years old, it is 0-1 year old with BT 250, FT 0 (rather than BT 250, FT 250) since the FT test does not push it back 250bars=1year further into the past. Recent data is considered better in the event that the "personality of the stock" is changeable in >1 year periods (for 250 bars settings) (ii) the BT usually invokes optimisation, either overtly or otherwise (eg the ASV test which cherry picks strategies based on their results) in the strategies to fix optimal settings (and choices of strategy) based on data in the BT period, whereas the FT uses those fixed settings in the same strategies when they process FT data. ie its not usually the same. You can tell this by looking at the % hit rate using performance reports on BT 250 and FT 250 over different date settings for "non-optimising" ie mechanical strategies You would anticipate BTHR% and FTHR% to be roughly equal always for so-called non optimising or mechanical systems but they seldom are in practice indicating that there is some difference occurring in test/use which introduces asymmetry in the results. Again my understanding which is empirical. best Richard | |
