# The Golden Cross, from Phil McDonnell

May 10, 2010 |

The Golden Cross has received a great deal of publicity. It is composed of two moving averages, the 50 day and 200 day respectively. When the 50 day crosses above the 200 day that is a buy signal. When the 50 crosses below the 200 day that says sell. Notably the 50 has crossed below the 200 in the turmoil last week.

As with everything this must be tested.

First I looked at the 200 day average alone for the daily Dow since1928. When the Dow closes below its 200 day MA then the return the next day is -.00665%. When above the 200 dma the Dow returns +.03968%. These compare to +.023% for all days. The p values for these two signals were each 7%. So they are not quite significant but you are not crazy if you still want to believe in the 200 day MA.

The question remains, what kind of improvement do we get when we add the 50 day crossover to the mix? On the sell side the 200 dma gave us an expected next day return of -.00665. But the 50 day golden cross sell signal gives a return of +.010%. It loses money!

The same thing happens on the buy side. The 200 dma returns +.03968 versus the golden cross return of .030%. In both cases adding the complexity of the 50 day average reduces return relative to just using the 200 dma alone.

## David Aronsen replies:

This is interesting Phil… I just want to be sure I understand what you tested. In the simple case using just the 200 day MA, did you look at the next day return for ALL day's whose close as < MA200 vs. ALL day's whose close was > MA200. Or did you look at only those days were the close moved across the MA200. I suspect the former as the number of crossing days would be very small in number.

With regard to the 50 and 200, I gather you looked at next day's returns for ALL days when the MA50 < MA200 vs. ALL days when MA50 > MA200. Yes?

As to the folks who say the golden cross is as valuable as gold (rather than the other thing) is that the returns from playing the crossings is what is of value, vs. let's say buying and holding or random signals with a similar frequency. Anybody know the data on that one?

`SELECT * FROM wp_comments WHERE comment_post_ID = '4738' AND comment_approved = '1' ORDER BY comment_date`