RichardL Veteran   Posts: 134 Joined: 9/18/2004  User Profile | Hi Adrian thanks for the very kind comments although to be honest I soaked up Jeff's books on Chart patterns and also some postings in the club forum. To tell you the truth, I am a bit jaundiced on the effectiveness of FT testing because: 1. on purely mechanical wholly non-optimised systems - you'd expect FT% to be roughly equal to BT% but it never is (looking at each strategy separately, ie outside All Systems Voting) 2. for good prospecting results, you need optimising systems which always skews to good BT and not very good FT typically - curve fitting at its best. But in any case, the FT test is not "for real" testing **unless you fix the backtest period absolute date range and don't use All Systems Voting **- you need true walk forward testing (THE most desirable OT future plug in EVER!) to give valid results that you'd hopefully experience in trading for real. Thats the only way I believe you can truly calibrate good strategies. Horribly time consuming to do manually, and just the thing OT could do automatically if one day configured to be able to On the questions (1) I like iTLB - any pattern in price or indicator/system is a good thing (for me) although have to be careful not to rely on connected underlying phenomena that therefore reinforce each other. I cull stocks purely because I don't think one strategy (as opposed to a superstrategy) can do all possible symbols - just not realistic. The usual number of internal parameters wouldn't be enough to fit the curves I think. (2) Confirmation, as I understand it is either you look for similar patterns in another symbol eg FTSE100 curve to confirm eg general rise for a suspected LONG opp on one of the FTSE100 symbols or Confirmation from other systems or indicators deriving from price/vol etc data on the SAME symbol. What we tried to do with the strategy, was to embody the eyeball confirmation tests we were applying to the Chart reported at the start of this thread but automatically, to present candidates which therefore already fitted the bill. cheers Richard [Edited by RichardL on 7/12/2005 6:28 PM] | |
