Yesterday Dr Rolf Wetzer flew in via Germany from his base in Basel to talk about his work with the Dax index. Past president of IFTA and editor of its journal, member of the Swiss Association of Market Technicians, he is a portfolio manager specialising in algorithmic trading. He delivered a condensed version of a presentation given at last year’s IFTA meeting in Sydney.
Quipping: ‘cycles are odd animals…but nobody uses them’ he explained how one starts from the premise that markets are non-linear and not stationary, which makes extracting cycles correctly rather difficult. He uses Empirical Mode Decomposition to search for cycles and from these creates oscillators. Based on the Hilbert-Huang transform and used by NASA in 1998, one can format almost unlimited Intrinsic Mode Functions which are like Algos applied to the data set.
Raw daily data is the first input, then one marks the interim high and low turning points to create an envelope, plotting its mid-point. One then adds a moving average (the length of which is an arbitrary decision) and from these we create oscillators depending on whether averages are above or below the mid-point. Repeat until you are happy with the result and you will find that the residual cannot be explained by cycles but is the intrinsic (and very subtle) long term trend.
Saying, ‘mathematicians don’t die, they just lose some of the functions that they have’ he likes the fact that this system provides a ‘reality check’ for traders as it shows the market without noise. It also provides suggested moving averages depending on the time horizon going forward and the amount of past data one ought to use.
If it all sounds a bit complicated and if you’re worried your maths isn’t up to boffin standards – don’t. Watch the video of the talk which will be out soon because he explains things well. And another thing: his system will provide hours of fun for the computer literate looking for the Holy Grail!
