On the web, there is a lot of material on how Full Moon and New Moon affect a variety of events on earth and the stock market is said to be no exception either.
Here are some links you may want to glance through,
Above and many others which can be easily searched for seem to suggest a linkage between Lunar Cycle and Market Returns. But as a trader, the question we need to get an answer for is
Does the linkage have predictive ability?
If we can see a Predictive ability in terms of Pre and Post Lunar day, we can then use that information to modify the style of trading to take advantage of the same.
For this test, I used BSE Sensex (since the history is the largest available among Indian Stock Indices).
Period of testing was from 1980 to 2013 (November)
Not being a statistical does limit the amount of tests I can run.
The first thing I wanted to do was compare and contrast (correlation in other words) the average returns of 5 days Prior to the D day to the 5 days Post the D Day.
Reason for choosing 5 (and not 4 or 6): A month consists of 30 days which is equal to 2 Lunar Cycles. In a 30 day month, we generally have markets open for around 20 days (Saturday / Sunday being holidays). 20 divided by 2 = 10 and hence I have used 5 for post information and 5 for Pre information.
While the time frame is long enough to dismiss Sample Size Bias, there is still a problem in terms of Lunar Days which fell on either Sunday / Saturday or a Holiday. Since the whole excercise was done in AmiBroker, it has omitted days when there was no data and hence the test may not be complete in all respects.
But since this can affect both New Moon and Full Moon, I believe that the probability of they cancelling each other is high and hence do not believe that they will affect my conclusions in a great manner.
Now for the raw results:
The Correlation between return (average) of 5 days prior to Full Moon compared to 5 days post Full Moon = 0.05
The Correlation between return (average) of 5 days prior to New Moon compared to 5 days post Full Moon = 0.01
The correlation being so close to Zero essentially suggests that returns are random in nature and there is no way to guess as to if markets will rise in the next 5 days if they have fallen 5 days prior to the D day.
Below is the other data points I thought may give some indication of the returns as well as the probabilities;
The only thing that caught my eye was that regardless of whether the previous 5 days were positive or negative, the next 5 were positive. But this can be easily explained by the fact that the markets have moved tremendously over the same period. The best way to remove that bias would have be to Zero Center the same, but that is for some day in the future 🙂