Thank you for your formula for the width of the bands, which shortens mine considerably. I haven't been able to come up with a good formula to gauge the width relative to the width over the stock's history. I'll take a look at what you did with StochRSI.
As I said, I tried these:
(mov((fml("Bollinger Band Upper")-fml("Bollinger Band Lower")),89,s)-.5)>(mov((fml("Bollinger Band Upper")-fml("Bollinger Band Lower")),200,s))
((fml("Bollinger Band Upper"))-(fml("Bollinger Band Lower"))>3.8)
mov((fml("Bollinger Band Upper")-fml("Bollinger Band Lower")),220,s)>(mov((fml("Bollinger Band Upper")-fml("Bollinger Band Lower")),520,s))
All are flawed (and could be shortened with your formula for the width of the bands). The problem is that if the moving averages are too short, you get buy and sell signals in periods of little volatility. If you make them too long, you cancel a lot of good signals. If you set a numerical value, it doesn't work for all stocks. The bands' width varies too much, depending on the price of the stocks.
The best gauge, of course, is the naked eye, but you can't work this into a profit testing system (not one that wasn't grisly, at least). |
