Trying to find a starting point for developing a first strategy I stumbled upon a lot of ideas – not at last on working through the manual - that I have ‘only’ to bring in the right sequence. However, I am a little bit confused about the role classical or traditional technical analysis plays within development of automatic trading systems. The zorro manual describes technical analysis generally as an example of ‘Trading methods that don’t work’ and declares traditional indicators as ‘useless for trade signals’, keeping a small back door open while ‘some [traditional indicators] can be often useful for auxiliary purposes’.
According to this I find for my taste a bit too much classical technical analysis within script examples or even in literature. The book ‘Trading Systems’ by Tomasini/Jaekle e.g. explains development of a trading system based on the LUXOR system, that Is a crossover / crossunder of two moving averages, and mentions in its appendix two further trading ideas: a Bollinger Band and a symmetric triangle system breakout. All three systems use IMO traditional technical analysis to generate trade signals.
In workshop 4 a simple trend trading system is shown where the trend should be identified by a LowPass-Filter. I don’t know the formula of this filter nor I can’t find it anywhere in the includes (am I blind?), but its sense should be to smoothen the price curve to identify turning points like many classical moving averages and even more complex classical indicators that consists of classical moving averages or mathematical derivations thereof try to. I can’t see the principle difference between this filter and classical moving averages and if or how the LowPass filter avoids the problem described in ‘what is an indicator’ in the manual on ‘using a transformation function from a n-dimensional space into a 1-dimensional space’. (Funnily enough, using a simple moving average in the example of workshop 4 increased the result in my back tests, but this only BTW and can also appear randomly in one special market – I have not pursued it further).

Surely it is possible that this phenomenon appears to me only as one because I’m still at the beginning of learning but the frequency traditional technical analysis crosses my way seems to be very strange if it is useless for generating trade signals.

Thanks, Sphin

You're right - technically, anything that produces a trade signal is an indicator. Therefore the manual distiguishes between the classical indicators used in traditional TA, and indicators that are based on scientific methods such as signal theory or statistics.

This is no sharp distinction though. I think a profitable system can be done even with classical indicators when you identify an inefficiency in a market and use a system of indicators that exploit that inefficiency. This would probably not work with a single indicator, but with a more complex combination.

In the book by Tomasini/Jaekle, I'd draw the border between the Luxor and Bollinger band systems that are based on classical indicators and are not really recommended for trading, and the triangle system that is based on the observation of an inefficiency and thus might have an edge.