carl » Fri Jan 20, 2017 6:56 am wrote:Hi Everyone,

I am not sure whether the implications of the method of calculating Slope are well understood. According to the code:
dblTma = iMA( NULL, tf, maperiod, 0, MODE_LWMA, PRICE_CLOSE, shiftWithoutSunday );
dblPrev = ( iMA( NULL, tf, maperiod, 0, MODE_LWMA, PRICE_CLOSE, shiftWithoutSunday + 1 ) * 231 + iClose( NULL, tf, shiftWithoutSunday ) * 20 ) / 251;
dblDiff = dblTma - dblPrev
Slope = dblDiff/ atr;

Now, it can easily be shown that the difference of the two LWMA moving averages is actually equivalent to:
dbfDiff = Close[0] - iMA( NULL, tf, maperiod, 0, MODE_SMA, PRICE_CLOSE, shiftWithoutSunday+1 );

What this means is that anybody who is trading the signals coming from the CSS crossing at zero, is in effect trading a signal which could simply be derived from observing the crossing of the Close of a simple moving average of the same period (shifted by one bar). This can easily be confirmed visually by adding the appropriate SMA to a chart and observe that the crossing of the SMA correspond to those of the CSS lines.

Carl

carl » Wed Jan 25, 2017 10:04 am wrote:Ray,
Here is an example (using the template of TraderJoeForex using M15, H1 and H4):
The three moving average periods are :
aqua - 7
yellow - 28
green - 112

This is a M15 chart.
The vertical lines show where the SMAs are crossed: The crossing times are
aqua - 0930
yellow - 1015
green - 1115

Only the aqua appears correct on all crossings. However the longer time frames appear incorrect (do not correspond to the ribbon). This is because the ribbon at the bottom is "belatedly" redrawn to correspond to the start of the timeframe of the indicator. Therefor on the 1 hour timeframe the boundary for 1015 is 1000, and on the 4 hour timeframe the boundary for 1115 is 0800.

This discrepancy highlights a problem with looking at historical signals generated with this indicator. The indicator then appears to call a turning point early whereas it may have actually occurred quite a bit later.
When using this indicator on a longer timeframe than that of the chart, it will probably be better to show the signal only at the time of the crossing, or alternatively at the end of the timeframe boundary.

Or, maybe just use the SMA?

Carl