I'm sure I am just overlooking something very simple and obvious, but I do wear reading glasses. I am working on a system test using both powerpivot retracement and trenddirection. Basically, I use trenddirection to determine if i can enter a long or short position on a minute chart using the next level up, ExtFml( "PowerPivots.TrendDirection",1 ,3 ). Whenever the trend changes direction, that is my enter long/short signal. Next, i want a decent sell, besides a change in direction. So I ran an optimization test with r1:=ExtFml( "PowerPivots.Retracement",1 ,opt1 , 3) trigger by a cross(r1,mov(c,5,s)). I ran this on LM, AHC, GOOG from 2/14/05 - 7/29/05. When i plotted and examined the data (GOOG) i was suprised to see that the timing of when the retracement flipped from under to over the price did not always, or often, match the trenddirection signal. The manual states for the retracement function, "It uses the TrendDirection function to first determine the trend of the market. If trendDirection is +1, indicating the market is moving up, it plats the retracement below the prices. If TrendDirection is -1, indicating the market is moving down, it plots the retracement above the prices." A simple test of this was as follows on a 1-min GOOG chart. One window has r1:=ExtFml( "PowerPivots.Retracement",1 ,.25 , 3); r2:= ExtFml( "PowerPivots.Retracement",1 ,.382 ,3); If(r1>r2,1,-1) (in blue) and a second plot in red of ExtFml( "PowerPivots.TrendDirection",1 ,3 ) On the price chart window i plotted r1:=ExtFml( "PowerPivots.Retracement",1 ,.25 , 3); r2:= ExtFml( "PowerPivots.Retracement",1 ,.382 ,3); r1;r2 and put r1 blue, r2 red to tell when they flipped. Again this showed the trend direction determination was different. Shouldn't they be the same? This wrecks total havoc on any system test depending on the cross of a retracement line. Thanks, JK