It seems you have JavaScript disabled.

Ummm.. Yeah... I'm going to have to ask you to turn Javascript back on... Yeah... Thanks.

The Golden Cross Trading Methodology at Trader’s Narrative

The Golden Cross Trading Methodology

Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

This is a guest post by Wayne Whaley, CTA:

I recently posted research on the 40 year (1970-2009) track record of various simple moving average crossover systems on the S&P Cash Index in a post titled “The 200 Day Moving Average Is Your Friend“. See that link for details, but a summary of those study results is shown below. I have added a new column, which shows the maximum loss in any given month for the various trading systems.

golden cross data table

To be clear, the 200 day system takes longs above the moving average and moves to 3 Month T-bills below the moving average.

The previous study deemed that the 200 day moving average system, where one moved to T-bills on all short trades was attractive because it’s return was competitive with buy and hold with much less risk than buy and hold (see “Largest Monthly Loss” column above).

There are almost an endless number of permutations of these simple moving average trading systems that can be studied. One that is popular is the “Golden Cross” system, where rather than receiving a signal when the end of day price penetrates the 200 day moving average, you must wait until a faster moving average (in this case, 50 day moving average) penetrates the longer (200 day) moving average.

I decided to look at 10-50 day moving average trends penetrating the 100-350 day moving averages, to see if the backtested results would improve the previous single day trading results. You can see the results below, with the same assumptions as before: go long when the shorter moving average exceeds 1% above the long moving average and short when the moving average exceeds 1% below the longer moving average.

Trading the “Golden Cross” system with the 50 day moving average against the 200 day moving average produced an annual return of 6.86%, with an average profit of 11.10% on longs and an average loss of -1.17% on the shorts. However the best system that I found over the 40 year database was to trade the 10 day moving average against the 250 day moving average.

This system resulted in an average annual return of 7.47%, with a 11.53% return on the longs, a modest -0.91% loss on shorts and it’s worse monthly loss was -14.45%. One possible drawback is that it would generate many more trades than the standard “Golden Cross” system.

golden cross data table

If you allowed yourself to move to Treasury bills on shorts, the Golden Cross 200/50 system would improve to 9.38% and the 250/10 system would improve to 9.71%

Summary of all of moving average crossover studies:

A Buy and Hold strategy over the last 40 years had an 8% average annual return and 11.5% when dividends were added.

The best single day price vs. moving average system that we studied was the 200 day moving average system with 9.20% return when short positions resided in T-bills.

The traditional “Golden Cross” 200/50 day system yielded a very similar 9.38% return when shorts were allowed to reside in T-bills.

The best system I tested over the last 40 years was the 250/10 system which yielded 9.71% when shorts were allowed to reside in T-bills.

The 250/10 system would have the drawback of increased transactions, that introduces additional cost and trading slippage into performance, which may offset its modest increased performance.

All of the systems discussed in above, had similar net performance to the buy and hold strategy, but much less risk profiles than the Buy and Hold strategy since they all spent roughly 1/3rd of the trading days in the security of T-bills.

Whether using the simple end of day price or some moving average of price, the results of this analysis supports the institutional popularity of the 200 day type moving average systems, with long, solid, consistent, low risk profit. However, there was little difference between trading performance between using end of the day price or a shorter moving average for generating the trading signals.

There is no guarantee that the market will show the same trading characteristics over the next 40 years that it did in the last 40 years and it is actually almost a given that a different combination of moving average systems will yield the optimum result going forward.

I suppose if you wanted to make a career out of studying these type systems, you could now look at the optimum exponential moving average to use as opposed to simple moving averages.

Enjoyed this? Don't miss the next one, grab the feed  or 

                               subscribe through email:  

6 Responses to “The Golden Cross Trading Methodology”  

  1. 1 DL

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    The study spanned 40 years. As we all know, markets react much more rapidly to news and data in the year 2010 than it did back in the 1970’s or even the 1980’s. So it is possible that data from the period 1970-1990 is skewing the conclusions one way or the other. I think it’s probably the case that at least one of the two moving averages that is used must be of shorter duration than would have been the case during the 1970’s.

  2. 2 wayne

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    Gosh Babak, as always, it read much better than the copy I sent to you a week ago.

    Before we get beat up for doing moving average studies, ad nauseum, I would like to pass along that the above study was a response to a comment by Mike C. inquiring as to what benefits would be gained by using a short moving average vs a long moving average (Golden Cross) rather than an end of day price vs a moving average. Since the code was basically in place to do that study, I cranked out the results and passed along to Mike and Babak, who felt it merited a post. I hope a few of you gathered some insight from it.

    Another comment I made to Mike, was that you could probably improve the results very modestly by using exponential moving averages, but the more parameters you enter into the trading system, the more prone the system is to being curve fitted to the time period tested. ‘Simple’ is usually better for forward results, if not necessarily past performance.

    Besides, rather than more moving average studies, I’m really itching to turn my calculator lose on the potential implications of these 4th quarter earnings numbers?

  3. 3 Kevin_in_GA

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    Great post - I had recently looked at EMA crossover systems very much like this, but only went back as far as 1995. I looked specifically at the 10 day EAM crossing either the 30, 40, 50, 60, or 70 day EMA.

    Long story short - the 10/50 cross was the most profitable yielding the following results:


    SINCE 1/3/1995 - $10,000 GREW INTO $45,953 (CAGR OF 10.7%) COMPARED TO $29,792 FOR B&H (CAGR OF 7.6%)

    SINCE 1/3/2000 - $10,000 GREW INTO $22,392 (CAGR OF 8.4%) COMPARED TO $8,699 FOR B&H (CAGR OF -1.4%)

    SINCE 1/3/2005 - $10,000 GREW INTO $20,239 (CAGR OF 15.1%) COMPARED TO $9,803 FOR B&H (CAGR OF -0.4%)

    SINCE 1/3/2007 - $10,000 GREW INTO $18,368 (CAGR OF 22.5%) COMPARED TO $8,093 FOR B&H (CAGR OF -6.8%)

    SINCE 1/2/2008 - $10,000 GREW INTO $20,779 (CAGR OF 44.2%) COMPARED TO $7,752 FOR B&H (CAGR OF -11.9%)

    SINCE 1/3/2009 - $10,000 GREW INTO $15,247 (CAGR OF 52.5%) COMPARED TO $11,801 FOR B&H (CAGR OF 18.0%)

    Now, between 1995 and 2000 it was nowhere near as good as buy and hold, but that period was an astounding run for the markets that would be essentially impossible to beat with any timing system. Truth is that the system really only kicked into high gear since 2005 which is why the recent data looks so strong.

    And remember, those CAGRs are based on the SPY only, not SSO/SDS …

  4. 4 wayne

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    Exponential Systems

    I ran the exponential moving average systems this morning over the last 40 years and they added no additional performance. Best long/short system was 6.33% annual return. If you cheat and assume the commodity has an upward bias and move to tbills on shorts, the best I could fine was 9.25%, in line with simple moving average systems.

    On to other endeavors.

  5. 5 Stefan

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    Here´s two charts showing S&P500 the last 90 years using monthly MA12 (similar to daily MA200). However, I prefer to use the slope of the moving average instead of looking at the index price compared to the MA. In some cases you get in/out a bit later but instead you avoid some bad trades when the price makes a false break of the MA.

    trendfilter monthly MA12
    trendfilter vs Buy-and-hold

    I have only used long positions, no short positions or T-bills during that time which would give an even better result.

  6. 6 MachineGhost

    Deprecated: preg_replace(): The /e modifier is deprecated, use preg_replace_callback instead in /home/traders/public_html/wp-includes/functions-formatting.php on line 76

    It would be far more useful to report results using an average of all rolling 12-month periods than the end of calendar years which is one form of curve fitting. Recall how Bill Miller of Legg Mason (before he blew up) was alleged to beat the S&P for like 15 years straight, but it was only when measured on a calendar year basis. Another example: monthly rebalancing is highly dependent on the starting/ending date, just one day more or less can turn what looks like a fantastically profitable system into an unprofitable one.

Leave a Reply