DEATH CROSS CREATES LIFE FOR BONDS

By Rom Badilla, CFA – Bondsquawk.com

A “Death Cross” is a technician’s term for a price graph pattern where the 50-day simple moving average fueled by a ferocious decline in prices, crosses below the 200-day simple moving average.  Technicians claim that this pattern confirms the change in trend where the price ultimately moves to a lower level.

The Chinese stock market, the Shanghai Composite Index, rallied significantly and led the global equity advance from the depths of the financial crisis.  The Shanghai Composite hit a low of 1664.92 in late October 2008.  After reaching that low, the Shanghai Composite index never looked back and rallied 109 percent to 3478.01 by early August 2009.  In a similar rebound, the U.S. markets bounced off the lows of 666.79 in March 2009, almost five months after the Shanghai reached its low, before hitting a high of 1217.07 on April 26, 2010 for an 83 percent gain.

After peaking followed by a subsequent decline, the Chinese stock market experienced earlier this year a “death cross” pattern.  After the convergence of the two simple moving averages, the Index has declined fulfilling the forecasts of technicians.  For the Shanghai Composite, the 50-day simple moving average crossed below the 200-day on March 1, 2010.  At that time, the Index closed at 3046.09.  Since then, the Index has dropped to 2419.42, a huge decline of more than 20 percent.

Shanghai Composite Index – “Death Cross” turned out to be Deadly

Last Friday, the U.S. equity markets caught up with its Asian counterpart as the two moving averages for the S&P 500 crossed.  While it remains to be seen if the recent “death cross” pattern for the S&P 500 will lead to lower prices (and possibly lower bond yields), it is worth looking back at history to see the consistency of this pattern in forecasting future returns in order to shed some light on future price movements for the equity markets.  Furthermore (and more importantly since our focus IS on the fixed income markets), we will examine corresponding yield changes once a “death cross” occurs in the equity markets.

S&P 500 Index – “Death Cross” Signal

Backtest of “Death Cross” Pattern for Stocks and Bonds

There have been 25 “death cross” patterns since 1962 and a total of 14 occurrences since 1980 for the S&P 500 (We do not include last Friday’s).  We note these two time periods since 1962 marks the beginning of available data for Treasury rates as provided by the Federal Reserve and 1980 more or less, marks the start of modern-day Fed policy.  If we track the performance of both stocks and bonds, three months after the “death cross”, we find that results have been mixed which suggests that the technical pattern is not a good predictor of price and yield declines.  Actually, it is the exact opposite just by this simple measure.

For the 25 occurrences dating back to 1962, the average price change for the S&P 500, three months after the “death cross” signal, is for an increase of 2.3 percent.  The biggest decline occurred in May 1962 when the S&P sank 12.5 percent, three months after the “death cross” signal.  The most recent, and second largest loss, happened in December 2007 when equities eventually declined 10.4 percent.  The biggest doom and gloom head-fake, where the “death cross” pattern appeared and equities rallied, occurred back in November 1986.  After the 50-Day crossed below the 200-Day moving average, the S&P jumped 20.6 percent three months later.

In addition, if we focus on the 14 “death cross” patterns since 1980, which again marks the beginning of modern day Federal Reserve Policies, the picture doesn’t change much.  The average price change for the S&P 500, three months after the formation of the pattern, stands at a gain of 4.7 percent.

Again, the average change over close to 50 years of data is hardly convincing evidence that a “death cross” ultimately leads to lower prices for stocks.

For bonds, the results are similar.  For the 25 “death cross” patterns since 1962, the average change for bond yields contrasts with what technicians may suggest.  Instead of seeing bond yields decline in response to a change toward a downward trend in stocks, we find that the average change is for an increase of 14 basis points, three months after the signal.  The largest increase in bond yields happened in July 1981, where the 10-Year Treasury jumped 189 basis points.  The second largest head-fake occurred in February 1984.  The 10-Year soared to 12.86 percent, an increase of 125 basis points, three months after the “death cross signal” in the S&P 500.  Comparatively, the largest decline in bond yields happened just recently, at the onset of the financial crisis in December 2007.  At that time, the “death cross” pattern appeared which was followed by the 10-Year dropping 83 basis points to 3.34 percent in the following three months.

Furthermore, if we just focus on the results after 1980, the evidence does not change much.  The average yield change for bonds in the 14 “death cross” occurrences stands at an increase of 6 basis points.

Below, Table 1 is Summary of Stocks, Bonds, and Fed Funds Rate.

New and Improved “Death Cross” Pattern

Now before we write off the effectiveness (or lack there of) of the “death cross” pattern, we should note that there are variations.  Specifically, in addition to the crossing of the two moving averages, the 200-day measure needs to be “rolling over”.  In other words, the 200-day when crossed by the 50-day, needs to be flat or declining in order to confirm a new downward trend in prices.  This variation better characterizes the most recent “death cross” pattern set last Friday.  So when we apply that parameter, obviously we have fewer observations to work with.  Interestingly, the results change depending on the asset class.

Since 1962, there are eight instances of this variation or enhanced “death cross” pattern.  The average price change actually increases for stocks, refuting the technician’s creed even further.  The average price change, over a three month period once the signal appears, is for a gain of 3.6 percent.

If we focus on the data just from 1980 where there are four instances of the enhanced signal, the average 3 month performance for stocks is 4.2 percent.  Performance after the “death cross” signal, reinforced with the “rolling over” 200-day moving average parameter, is similar to the aforementioned average of 4.7 percent derived from the simple signal.

Interestingly, bonds reveal a different story.  Since 1962 and in the eight instances of the enhanced “death cross” pattern, bonds on average rallied as yields fell.  Five out of the eight signals led to a decline in Treasury rates as the average of the eight reflects a drop of 14 basis points.

For the four results post-1980 which again represents the onset of modern-day Fed policies, the average decline jumps to 56 basis points.  Furthermore, out of the four signals, all four reflected a decline of Treasury yields over a three month span.

Now it would be imprudent for me to not come up with a counter-argument for this case since one can easily argue that post-1980, there was a secular decline in interest rates.  So, owning bonds from that point on, whether there is a “death cross” signal or not, would have undoubtedly resulted in positive returns.  Indeed, such a stance has merit since it was PIMCO’s model to own and hold longer maturity bonds during that time frame.  This outperformance eventually led them to becoming the largest bond portfolio in the world.

In an effort to refute I will point out that in each of the four instances, the Federal Reserve lowered the Fed Funds target rate suggesting that it was more than chance or the aforementioned secular decline that led to a drop in rates.

For the “death cross” signal in April 1980, the Fed Funds target rate dropped 200 basis points in a three month span.  For November 1987, short term rates declined 25 basis points by the time February 1988 rolled around.  The Fed lowered the target rate 50 basis points following both the September 1990 and October 2000 signal.

Below, Table 2 is a Summary of Bonds, Stocks, and the Fed Funds associated with just the enhanced “death cross” pattern.

With these four instances and given that stocks are a forward looking mechanism on the economy, we can surmise that the “death cross” patterns occurred as a result of weakening fundamental data on the economy.  These same fundamental signals ultimately led the Federal Reserve to act by lowering borrowing rates, which in turn led to lower yields and outperformance for the bond market.  In response to the changing landscape of lower borrowing costs, which is beneficial for equities, stocks rebounded.

At the end of the day, the technical “death cross” pattern by itself, using either the simple or enhanced method, has mixed results in terms of projecting equity performance.  Ironically, bond bulls have to look to the equity markets to reinforce their position.  Keep in mind that our judgment of this technical signal is based off of a set time frame which is three months.  Since we are using a technical indicator to imply an initiation of action, it only makes sense to use technical indicators to exit out of it.  Admittedly, our method of a three month window offers limited insight on its true effectiveness since technicians typically use price objectives, which are determined by other analysis such as Fibonacci retracement levels and Stochastics in order to come up with an exit strategy.  Hence in a “don’t try this at home” warning, judgment should be reserved on the “death cross” technical signal.

Having said that, the “death cross” signal as the name implies, should not be taken lightly.  After all, it is my belief that price is the true arbiter of value for any investment and such a signal should provide context for the overall fundamental picture.  Meaning, if stocks are declining and the 10-Year is trading sub-three percent which results in a signal, it is happening for a reason.  It is up to the investor, to determine what that is exactly and whether it refutes or reinforces their current investment thesis.  As for the most recent ”death cross” established last Friday, it should be fairly obvious what the signal suggests given the backdrop of high unemployment, declining price pressures in real estate, and tumbling inflation expectations, coupled with the debt crisis in both Europe and stateside.

BondSquawk

BondSquawk is written by a team of bond market experts whose aim is to provide an unbiased view of one of the largest (but under reported asset classes in the world) – The world of bonds.

More Posts - Website

  • Big Mac

    Rom B – Thank you for your detailed and well documented analysis. I believe that a 50-day simple moving average crossing a 200-day is more a long term warning of potentially big trouble ahead rather than a good indicator of what next quarter will bring. Additionally, the non productive signals this indicator generates tend to be short and shallow. The most natural endpoint for measuring the complete effectiveness of a death cross should be where the market is when the 50 day recrosses (golden cross). For example, the 12/21/07 cross did not reverse until 6/23/09 after a 589 point plunge. On the other hand, the 07/19/06 “fake out” ended on 8/29/06 (6 weeks in duration and 45 points of potential upside lost). This is clearly a different picture than the 3 month end point used (-10.4% versus + 8.3%) presents. I would encourage you to look at the full impact (cross to recross) if you want to truly understand how effective this indicator can be for preserving capital…..Thanks…….Big Mac

  • billw

    Rom,

    Really well documented data, and I agree that this indicator should be viewed in context with all other relevant data for a given time. In light of that I agree that the fundamentals this time are saying that this death cross is correctly signaling a near term major market correction. The downward trend in the GDP alone will probably send this market heading to the tank soon.

  • walden

    Since the market is up ~4% the last two days, it’s already made, on average, what might be predicted for the next 3 months.

  • BobbyP

    Anyone can run a chart of the S&P 500 with Simple Moving Averages (SMA) over the past ten years and see that the 50 and 200 day SMA gives false signals. I’d like to see the author make the same argument using the 50 and 200 day Exponential Moving Averages (EMA). Choosing the SMA for the analysis is a straw argument. The 50 and 200 day EMA crossover can also give rare false signals, but the author would have a much more difficult time making this argument. Anyone with a reasonable understanding of technical analysis would know this and would not have used the SMA for this analysis in my opinion.