To: TA2K who wrote (2895 ) 3/8/1999 7:35:00 PM From: Craig DeHaan Read Replies (1) | Respond to of 4056
Nicho*, Sounds like a valid theory execution. That one'll work; you could abbreviate it to - Close > Ref(HHV(Close,90),-1) or use Close = HHV(Close,90) and/or stage increasing longer lookback periods against shorter ones to see if the hills and dales are net rising or falling. Or you could experiment by setting % peak / trough value benchmarks on that topography in the same way so you don't limit yourself to an arbitrary or specific HHV lookback period. Something that would look for price surpassing a 10% prev price peak (or you could adapt it to any specific indicator value) might be - Close > Ref(peak(1,Close,10),-1) OR Close > peak(2,Close,10) Tho peak / trough functions use zig-zag to determine existence, once you move beyond the most recent peak/trough ocurrences (meaning 2nd or 3rd most recent), they are more likely to be "fixed" and therefore slightly more believeable. Here's an example of the pitfalls of Z-Z. Test this system that extracts the worst of the peak / trough delusions and you'll learn thru actual experience. First set up a custom indicator called "Price OscPeak" - If(PeakBars(1,CLOSE,2) < TroughBars(1,CLOSE,2), {then}CLOSE-Peak(1,CLOSE,2),{else}CLOSE-Trough(1,CLOSE,2)); Then test this system - ::Price Osc Peak System Test:: <<enter long>> Blong:= If(Cross(Fml("Price OscPeak"),0) AND (Ref(Fml("Price OscPeak"),-2)<Fml("Price OscPeak")),1,0); Blong=1; <<enter short>> SShort:= If(Cross(0,Fml("Price OscPeak")) AND (Ref(Fml("Price OscPeak"),-2)>Fml("Price OscPeak")),1,0); SShort=1; Definitely don't count on the results to make next months mortgage, or prepay that new MBenz. <g> CP