Monday, 25 February 2013

Gold’s death cross is no reason to feel grim

Commentary: Predictive nature of indicator has been weak lately

By J.J. Zhang
J.J. Zhang is chemical engineer and amateur financial adviser who was the winner in MarketWatch’s second annual World’s Next Great Investing Columnist contest. He runs the blog MarketTech Reports. You can follow him on Twitter @MarketTechRpts.
Technical price and momentum indicators are popular tools for many investors as a way to decide when to enter or exit positions. Recently a popular indicator has been making the rounds on most financial sites due to gold hitting the “death cross.” So what is a death cross?
Best of MarketWatch
Will health reform make your insurance cheaper?  
Here's the one other story that  you can't afford to miss today
The definition of a death cross can vary depending on who you ask but generally it refers to the 50-day moving average crossing under the 200-day moving average. A common condition added to this definition is that the 200-day moving average must also be on a downward trend.
Generally this is seen as a bearish sign and is supposed to indicate that the market has a negative outlook on the future of gold prices.
So with gold entering a death cross, you have to ask: how useful is this indicator? While many technicians do use other indicators in combination with the death cross to get a better feel of market direction, it’s still an interesting statistical exercise to look through historical data.
After all, how often do death crosses and their opposite, the golden crosses, occur? What is the typical price performance a week, a month, a quarter and a half year after? Do they actually predict longer term market deterioration?
The first question of death cross frequency is a relatively simple one to answer using historical data. Using gold spot price available from the World Gold Council at Gold.org, an analysis found the commodity metal has seen 37 total 50/200-day moving average crossovers (including last week) since 1979.
That works out to 19 death crosses and 18 golden crosses at roughly 1 crossover per year.
The tables below list each death and golden cross occurrence along with the number of trading days each period and the percent change after one week, one month, three months, six months and until the crossover reverses (by definition, a golden cross always occurs after each death cross and vice versa).

Death cross performance since 1979

Death cross Trading days lasted % change after 1 week % change after 1 month % change after 3 months % change after 6 months % change ar reversal
Feb. 20, 2013 TBD TBD TBD TBD TBD TBD
April 13,2012 113 -1.5 -6.5 -4.3 4.2* 6.2
Feb. 10, 2012 10 0.7 -0.8* -6.6* -5.4* 3.8
Sept. 1, 2008 116 -1.7 7.0 -5.4 14.0* 8.9
Oct. 19, 2006 43 -0.2 4.7 5.4* 14.2* 2.8
May 30, 2005 56 1.8 4.5 3.0* 18.5* 6.0
June 4, 2004 85 -0.9 2.6 3.4 16.7* 7.0
May 3, 2000 290 0.7 2.8 -0.1 -4.0 -2.1
Dec. 14, 1998 221 -0.6 -2.0 -0.6 -10.7 6.9
June 12, 1996 642 0.4 -0.3 -0.7 -4.2 -23.0
Sept. 27, 1995 66 -0.3 -0.3 0.9 4.3* 0.9
Dec. 16, 1994 90 0.3 -0.7 0.9 2.9* 3.4
Oct. 2, 1992 155 0.6 -2.4 -5.7 -2.3 2.2
Jan. 30, 1992 157 0.0 -1.6 -5.5 0.2 -3.9
Feb. 27, 1991 202 1.7 -1.7 -1.2 -1.7 1.3
May 1, 1990 104 0.3 -1.6 0.3 2.9* 5.4
March 2, 1988 452 1.8 6.5 6.5 0,.0 -4.4
Aprl 29, 1983 588 0.5 1.9 -1.7 -11.0 -24.2
Jan. 7, 1981 432 -0.6 -10.8 -12.5 -30.8 -20.8
Avg.-total 212 0.2 0.1 -1.3 0.4 -1.3
Median-total 136 0.3 -0.5 -0.7 0.1 2.5
Avg. last 10 years 71 -0.3 1.9 -0.8 10.4 5.8
Median last 10 years 71 -0.6 3,.6 -0.7 14.1 6.1
* indicates after reversal
For death crosses, the average length is 212 trading days while the median length is only 135, indicating a few occurrences’ lasted unusually long. Within the last 10 years, the average has only been 71 days, a significantly shorter length.
In fact, none have lasted six months or longer since 2000. If the pattern holds, the current death cross may only last 3 months. 


Continued from page 1
Page 1 Page 2
Surprisingly, the average change after a death cross at the one-week and one-month marker is just barely positive at 0.2% and 0.1% respectively. The three-month performance shows a small drop at -1.3% while the six-month performance is positive at 0.4%.
However, this result is driven heavily by the last 10 years. Because each death cross lasted for such a short period, most of the three- and six-month data include a significant amount of time where gold was in a golden cross bull market.
Pre-2004, gold generally saw a long-term deterioration in market price after a death cross, averaging -4.5% after six months. In comparison, post-2004, gold averaged a 10% gain six months after a death cross and 6% by the time the cross had reversed.
This indicates recent death crosses don’t signal much regarding the longer term market outlook for gold. While continuing to assume this trend can be dangerous, nonetheless, it shows gold’s relative robustness in today’s market.

Golden cross performance since 1979

Golden cross Trading days lasted % change after 1 week % change after 1 month % change after 3 months % change after 6 months % change at reversal
Sept. 19, 2012 111 -1.3 -1.7 -5.8 TBD -9.0
Feb. 14, 2012 35 -4.0 -5.5 -11.8* -6.2* -6.1
Feb 10, 2009 783 6.4 -0.9 0.4 3.6 92.0
Dec. 19, 2006 444 0.0 -0.9 0.3 0.1 -0.6
Aug. 16, 2005 307 -0.8 3.2 7.4 21.6 34.1
Oct. 1, 2004 171 0.9 2.6 4.2 2.2 0.1
June 13, 20001 777 -0.2 -2.1 3.1 0.9 43.5
Oct. 19, 1999 141 -4.2 -5.1 -7.1 -9.1 -11.0
Nov. 27, 1998 11 -1.3 -3.5* -3.2* -9.2* -1.8
Dec. 28, 1995 119 1.9 4.5 3.0 -1.3* -0.5
April 21, 1995 113 -0.2 -2.3 -1.1 -2.0* -1.7
May 7, 1993 420 3.0 4.3 6.8 5.1 6.1
Sept. 7, 1992 19 1.0 2.1* -2.4* -4.5* 1.6
Dec. 6, 1991 39 -2.4 -5.0 -5.0* -7.9* -3.5
Sept. 24, 1990 112 1.0 -5.5 -2.2 -8.5* -8.5
Nov. 24, 1989 113 -0.5 -0.7 -1.1 -11.9* -11.6
July 31, 1985 675 -1.9 2.2 -0.7 7.0 31.1
Sept. 6, 1982 169 -7.4 -14.7 -4.1 -11.2 -9.6
Avg. --total 253 -0.6 -1.6 -1.1 -1.8 8.0
Median- total 130 -0.4 -1.3 -1.1 -2.0 -1.2
Avg. - last 10 years 309 0.2 -0.5 -0.9 4.3 18.4
Median- last 10 years 239 -0.4 -0.9 0.4 2.2 -0.3
* indicates after reversal
For golden crosses, the average and median length is similar at 253 and 130 days. However, the last 10 years averaged a length of 308 trading days, showing just how strong the recent gold bull market has been.
Similar to how the one-week and one-month performance counter-intuitively went up after a death cross, average one-week and one-month performance dropped after a golden cross at -0.6% and -1.6% respectively.
Surprisingly, average three-month and six-month performances were also negative after a golden cross at -1.1% and -1.8%, contradicting the popular notion that golden crosses signal a bull market. However, due to the long period of each crossover event, gold eventually returned 8% on average by the time a golden cross reversed into a death cross.
Like death crosses, the last 10 years for golden crosses also show a significant deviation versus pre-2004 trends, instead displaying a strong positive gain at six-months.
Overall, a simple analysis of death and golden crosses since 1979, ignoring the more nuanced interpretations and other technical indicators, indicate relatively poor long-term market predictability for gold. While death crosses were relatively good indicators pre-2004, recent death crosses generally are short lived and prices recovered soon after.
Golden crosses are also an overall poor indicator though the long gold bull market in recent years has helped the correlation. Whether that counts as a predictive indicator is debatable however.
In the end, while death and golden crosses sound important; don’t get too worried about them — there’s much more in the financial markets to be concerned about.
Note: For more statistics and graphs from this study, please visit my blog, linked above.
Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)