Beginner here, so apologies for the rather naïve question, I try to find answers elsewhere but mostly I found opinions (often contradicting and rarely with an explanantion). Maybe I'm using the wrong terminology in my searches, or maybe it's such a basic concept that it's not discussed... Anyhow..
I started exploring the quant world recently and I would like to apply some automated trading strategies to long term investments (i.e. trades on a weekly basis rather than daily or intra-day), so I tend to download and visualize data using weekly aggregation (i.e. 1 candle/1 HLOC = 1 week)
Many technical indicators have been optimized for daily data and their parameters are the result of that optimization (e.g. 200d for trend using SMA, 20d/2SD for Bollinger bands, 12d/26d/9d for MACD, etc.)
SO the key questions I have are:
- Should I be scaling the parameter to fit the data while preserving the original time reference? Using the examples above, 200d SMA -> 40w SMA, Bollinger bands 4w/2SD, etc
- Should non-temporal parameter such as SD be adjusted too? According to Bollinger himself, changing the period should lead so adjustments of the SD
For consistent price containment: If the average is lengthened the number of standard deviations needs to be increased; from 2 at 20 periods, to 2.1 at 50 periods. Likewise, if the average is shortened the number of standard deviations should be reduced; from 2 at 20 periods, to 1.9 at 10 periods.
Although he also states
Bollinger Bands can be used on bars of any length, 5 minutes, one hour, daily, weekly, etc. The key is that the bars must contain enough activity to give a robust picture of the price-formation mechanism at work.
So I assume I can keep SD at 2
- What about the MACD? The Moving Averages period are not multiples of 5, 12d/26d -> 2.4w/5.2w, do I round them? What about the "signal"? That is derived from the MACD line, should that be scaled too?
Any guidance would be greatly appreciated