My NDX Mom system is simple and delivers good backtest results (CAR > 30% over 10+ years). However, looking at the chart of some buy candidates makes me feel uncomfortable. E.g. on 31/7, the system suggested NFLX. Still very high on the rank list (ROC based) but with the recent price drop, the chart is not looking very promising anymore (from my point of view).
Hence, I went back to the drawing board looking for an additional criteria which would remove candidates with a high risk of entering a downtrend (or just moving sideways). I would expect this could further smooth the equity curve and reduce max DD and the number of loss months (apprx. 25%).
I played with several criteria which helped on some of my other systems (e.g. rank = ROC/ATR, gap filter, ROC(short term ) > x or ATR(x)/MA(C,x) > y) .
I also tried trendlines:
y0=LastValue(Trough(C,5,2));
y1=LastValue(Trough(C,5,1));
x0=BarCount – 1 – LastValue(TroughBars(C,5,2));
x1=BarCount – 1 – LastValue(TroughBars(C,5,1));
Line = LineArray( x0, y0, x1, y1, 1 );
Cond = y1 > y0;
The trendline approach appeared quite logical to me in this context. Unfortunately, backtest did no show any improvements.
So, if anybody has an idea which criteria could work would be greatly appreciated.
