Ask Colin

With the three moving averages, you mention those used by Dawn Bolton Smith. Which exponential averages have given you the line of best fit and therefore which do you use?

If I understand Dawn's work correctly, she always uses the harmonic set of moving averages 5 days, 15 days and 30 days. I assume she would also use 5 weeks, 15 weeks and 30 weeks in longer-term work, though I cannot recall seeing her show a weekly chart.

I also believe that Dawn only uses simple moving averages, whereas I sometimes use exponential moving averages. The latter are useful where there are unusual market movements that may bias the averages as the data \d\r\o\ps out of the time window (refer Dr Alexander Elder's book Trading for a Living, pages 120 to 123).

I am not sure what you mean by the "line of best fit" for a moving average or a set of moving averages. This is a term usually associated with regression lines, or what some people call internal trend lines, which is a quite different technique.

If you simply mean which set of moving averages seem to work best, I am not sure I can answer you based on any rigorous work. For one thing, I think trying to optimise moving averages is dangerous stuff, because it may lead you quickly into the error of curve fitting to a particular set of data. There are ways to optimise safely, but to me the whole thing is an exercise in futility.

What we want is a set of averages that work well and will still work if we alter their lengths somewhat either side of the chosen periods. This makes them more robust and they should work in varying markets. I am quite willing to look at alternatives, but I find Dawn's set work as well as any. She has used them in a variety of markets and over many years and that she still uses them successfully speaks for itself.