Tuesday, November 3, 2009

Fractal Finance

Ever heard of the Elliot Wave Theory? Maybe you have, but I hadn't until last week. Put simply, this theory suggests that markets behave is a predictable way which is not driven by fundamentals (actual production of goods, actual jobs, etc) but simply by human behaviour in the marketplace – the collective investor psychology.

The image below show the fundamental Elliot wave – 3 peaks (1, 3, 5) and two troughs (2, 4) on the way up, and two troughs (A, C) and one peak (B) on the way down.

While quirky (as an economist I like to think in terms of the fundamental patterns of production) this theory has a lot going for it.

1.  It is fractal and scaleless. That is, the pattern reoccurs over very small and very large time periods.
2.  The theory enables bulls and bears to be correct much of the time. It goes up and down.
3.  It seems irrational and predicable – a good reflection of human behaviour.

However there are some problems.

1.  It is impossible to be wrong if you simply say the market will go up and down (the most basic claim of this thoery).
2.  The theory has nothing to say about the speed and scale of each wave; it therefore does not add to the forecasting arsenal.
3.  It assumes that markets ignore all political and technological change, or it assumes these things change in a wave pattern.
4.  It assumes that markets actually measure something. Should Elliot wave analysts use prices in real terms or nominal terms?

So where are we on the basic wave formation? Did we just have a 3-4 crash, or a 5-A crash?

I have a chart of the All Ordinaries index from 1985 with my amateur attempt at using Elliot waves to forecast the next market move. Umm… down maybe?

While interesting, how anyone uses actually applies this theory to reveal anything useful escapes me. You can read up on Elliot wave analysis of the current market here, or do a google search for commentary by Robert Prechter, the current guru of Elliot waves.


  1. Hmm... If it looks like pseudo-science, walks like pseudo-science and talks like pseudo-science then it probably is pseudo-science...

    A theory that explains everything is useless. I can't see how you could ever falsify this model?

  2. Hi Cameron,

    My apologies for the heat on the BS blog. I see you are still continueing to post, and I like this wave theory. You also seem to have an enviromentalist bent, but from a practical view and not tree hugging for the sake of hugging, so that is a plus.

    I did post on Andrew Leigh's blog but it disappeared. Not sure if that was gremlins, but I suspect my 3 questions were too hard or too questioning of his claim that the cost of stamp duty does not affect prices.

    I have a related question. When did economists begin to believe that their field was a science? I must have missed that transitionary period. Do you think that the complex maths and engineering principals used nowadays leads economists to think that conclusions drawn without taking human behaviour and pyschology into account will hold true regardless?

    I am puzzled by the self confidence shown by many who clearly have little experience other than a purely academic one.

  3. Thanks for the comment.

    Umm.. depending on how you define science economics could fit the definition. Just like astronomy and meteorology though, you can’t undertake experiments and thus test an one outcome against another under controlled conditions. An emerging field of experimental economics is trying to tackle some of these problems in simple decision making settings, and many people look to natural experiments, accidents of history, that provide experiment-like data. For example, in countries where some regions apply different tax treatments, regulations or laws, but are otherwise almost identical, you can look to the before and after changes in the control and test case to estimate the effect of the tax/law etc.

    The book The Myth of the Rational Market discusses the emergence of quantitative analysis, and the subsequent increasing confidence in numerical forecasts throughout the 1950s and 1960s.

    Actually, I personally hate the maths and model obsession. Where you start with a solid theory and use maths in an attempt to quantify the scale of the outcome it is generally adds to the discussion. Or conversely, use maths to search for trends in the data, and see if that fits existing theory, or helps develop new theories. Often times though, it simply adds confusion (I have an October post about property development lobbyist misusing data and quantitative techniques).

    I like Leigh’s outcome because traditional economic theory would predict exactly what he found - his method is simple and he a quite decent data set.

    Also, economists are essential behavioural scientists, and indeed the emergence of behavioural economists as a disciple that can explain why humans don’t always respond in the same way to incentives, is filling many gaps in knowledge.

    Finally, economists have shown that experts are more confident with their predictions than lay people in many fields, even though their predictions are no better.

    Until next time.

  4. I'm ok with that. We have both used up our powder on the stamp duty debate, so lets put that to bed.

    Best of luck.