LunaticTrader

Investing with the Moon

Questioning financial astrology – part 1

Posted by Danny on October 10, 2013

From time to time I get questions whether I use other planetary cycles besides the moon. Astrologers tell me to watch out for the next opposition of Jupiter to Saturn, or warn me that Uranus will be going retrograde next week. What about the Bradley indicator? Or the solar cycle?

So let’s have a look at financial astrology and why I don’t use most of it.

The problem with most planetary cycles is a simple one: we don’t have enough stock market history to reliably test any cycles that are longer than a year. For example, the planet Jupiter takes about 11 years to complete one cycle around the sun. Since the start of the New York Stock Exchange the planet Jupiter has completed only about 20 cycles. How much statistical confidence can we have in whatever conclusions we might derive from these 20 observed cycles?
Maybe stocks have gone up when Jupiter was in Aries 14 times out of the observed 20. Should we then bet on the market going up again the next time Jupiter enters Aries?
Well, there are 12 different star signs, so it is completely normal that stocks will have been a bit more “lucky” when Jupiter was in one of those signs (could be in Aries or any other one). So, the answer is no because this can easily be the product of random chance.
Imagine you see a person toss a coin 10 times and get tails 7 times. Would you then bet that he will obtain tails again on his next try?

The statistical significance is just too weak when we have 20 or less observed cases.

But that’s not the only problem. Let’s say we continue to search and find some cycle that matches the stock market history in a way that is very unlikely to be the product of random variation, maybe at a 99% confidence level (see: p-value). That’s something we can use, right?
Well, not so fast.
We have now run into the “look-elsewhere effect“. We looked in all the different planets, we looked in combined cycles like Jupiter-Saturn conjunction cycle, we looked into different aspects, we looked into retrograde motion and heliocentric cycles, we looked into planetary nodes, and so on… until we found something that “worked”.
But if we look in a few hundred methods and cycles of different lengths it is completely normal to find a few that stand out at the 99% confidence level.
This is no different from continuing to toss coins until you manage to produce heads 7 times in a row. It will happen sooner or later.

So, apart from the problem that we don’t have enough market history to test most planetary cycles for stock market effects, we also have so many different astronomical cycles and astrological methods to test that we are guaranteed to find a few that seem to work very well over a given test period for a given financial instrument. Are we then looking at the hand of luck from the look-elsewhere effect, or is it a genuine market cycle?

In a second part I will present another take on long term planetary cycles, and a possible use.

Stay tuned,

Danny

Enhanced by Zemanta

8 Responses to “Questioning financial astrology – part 1”

  1. Robert said

    Great post!

  2. Gaylord Rohloff said

    Hi Danny, Interesting, though I’m already in the choir. I would point out, with 100% accuracy, that every time Pluto is in Aquarius, New York starts a new stock exchange. Gaylord

    • Interesting comment. But as a (very ordinary) punter, rather than one of the congoscenti, I’m not certain of its significance ;-)

      How often is Pluto in Aquarius?

      • Danny said

        Pluto takes 248 years for one cycle, so the last time it was in Aquarius was about when the NYSE was started, as Gaylord jokingly points out.
        Pluto is the slowest cycle, but also Neptune (165 years), Uranus (84 years) and Saturn (29 years) are very slow, so only a few of their cycles have been observed since we have stock markets.

  3. The bias decreases monotonically with the number of variables used, in other words I can perfectly predict the past by using a sufficiently large number of random variables that would have clearly no power to predict the future.
    Did you try to test the lunar cycles on DJIA back to 1885?

    • Danny said

      Hi Fabrizio,
      The basic testing was done on S&P500 since 1950. That gives us over 700 months, which is a reasonably large sample to look for lunar effects and formulate hypotheses. Extending it further back to the 1800s would be of little use, because as you point out, curve-fitting is always possible given a set of random variables. The only true test is in forward testing. We started forward testing of the main lunar cycle hypothesis when this blog was started in summer of 2009. The results are kept on the “Performance” page: https://lunatictrader.wordpress.com/performance/
      The lunar effect has continued to show up in the last 5 years.

      Danny

      • Hi Danny,

        There is a reason I am asking you the question that would be too long to post on this blog. Do you have an email address to contact you so we can discuss about it?
        Mine: fabrizio.maccallini@gmail.com.

      • Danny said

        Hi Fabrizio,

        Feel welcome to post it here. Maybe other readers will benefit/comment as well.
        As far as I know the WordPress system can handle long comments, and you can also split it up in more than one comment if you want.

        Danny

Post a comment (disagreement also welcome):

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

 
%d bloggers like this: