Market Timing: Should You Attempt It?

Top 50. Oct 1, 2000 9:24 PM

» JenL_2 - Testing Your Metal

Hi Michael - Welcome to Suite 101!

Well can't say that I've been through a Real Bear Market. I was blissfully unaware of the 70s bear market - being out of the country as a Peace Corps Volunteer with no home ownership, no investments, and college loans to pay off. There's something to be said for having nothing to lose.

But I did go through the 1987 market crash. Had only been in my company 401(k) plan for about a year, 100% in Fidelity Magellan. As I remember, the loss seemed huge at the time, but I continued to DCA in my 401(k). In the meantime my company 401(k) plan changed, and I went 50/50 into Vanguard's S&P500 and Primecap funds, still DCAing. In about a year my 401(k) account recovered the loss, and then went up from there.

Also continued to DCA throughout the flat market in 1994. I considered it a feather-nesting year in which we'd continue buying shares in mutual funds in the 401(k) plans and IRAs, and then wait till the market turned up again and our shares "hatched", and indeed they did with the next leg-up of the bull market.

And as Marty recalled, the Correction of 1998 was another test of the metal of an investor. A lot of us here went through that correction together, as this discussion site became more like a support group with our mantra....

"WAGS = We Ain't Gonna Sell".

So no - guess you could say most of us haven't been through a Real Bear, but I think that most of us have been through enough gut-wrenching market zigs and zags to know if we have the where-with-all to stick one out. And if there is any doubt, then we shouldn't be in the market.....Jen

-- posted by JenL_2

Top 51. Oct 1, 2000 10:08 PM

» Kirk - WAGS!

Yes, we have a great support group here. Many of us were buying at or near the bottom in '98 and made a ton. Not EVERYONE sells, despite what some will say.

I get a real kick listening to timers and "experts" telling us what "everyone" will do when often these very same people are such experts because THEY were the ones selling in the last bear market!

I held stock through the 1973/74 bear (2o shares of Safeway while I was in high school) and I got to learn why people like dividend stocks. In the 1987 bear, my HWP stock was halved and took 4 years to break old records. I kept buying all along and didn't sell any. My mutual fund manager at Vanguard Windsor for my 401K sold at the bottom and I wrote a VERY nasty letter to HWP management. Many years later they changed the whole 401K program and returned my letter and asked if I was pleased with the changes. Later on in the early 1990's, I learned of Brinker and one of my first questions I asked when everyone told me how wonderful he did was "did he hold or sell in 1987?" I learned he sold so he failed my test from the get go. I subscribed to a newsletter by Jay Schabacker in the early 1990's but dropped it after doing better picking funds on my own (he's done 8.5% annuallized since 1990 according to Hulbert!).

Anyway, I just read an article that said it is the individual investors that are the best at buy and hold or buy the dips and it is the lemming mentalitly of fund managers who give up performance to reduce risk of coming in last. That means they sell into weakness often just so they don't get stuck with a stock when it is at the bottom. They don't get fired for being average, but they do for being last.

-- posted by Kirk

Top 52. Oct 2, 2000 6:34 PM

» michael_h - Once Again

My main point in my post was that neither buy-and-hold nor timing is a panacea. Looking back at market history in this century, there have been long periods when equities went nowhere. The returns we have seen over the past 18 years are abnormal. So are the short durations of bear markets and the quick recoveries. Of course, maybe things have changed...but we won't know that until we are looking back at this period with hindsight.

Regarding 1973/74, there is nothing that says exogenous events cannot occur which will bring a worldwide recession of great depth. Sentiments such as "these times are different" were also voiced in the Roaring 20's (just before the Great Depression) and in the Go-Go 60's (just before the awful times of 1968-1982). Perhaps we are due? Fourteen years of zero returns in the major indices (such as happened not that long ago) will cause many opinions to change. If you are a confirmed buy-and-holder, I hope you pay attention to what the Finance professors preach about asset allocation. Are your investments spread across a diverse group of asset classes, growth stocks, value stocks, small-caps, international stocks, various types of bonds) or are yours clustered in growth funds, tech funds, the S&P 500 Index, etc.? If so, you are at risk! Pray a real bear does not visit Wall Street!

-- posted by michael_h

Top 53. Oct 2, 2000 6:39 PM

» michael_h - Thanks

Jen:

Thanks for the warm welcome! Investing can be psychologically taxing. I think the support buy-and-holders get on this site is a very important support mechanism. And given the upward bias ofthe market and given enough time, buy-and-hold always works! Thanks again for the welcome.

-- posted by michael_h

Top 54. Oct 2, 2000 8:16 PM

» oneputt - bear market psychology

I agree with a well known radio personality, that I would never want to ride a bear market down. This seems obvious, who would really like to loose money when they could avoid it. The REALITY of the situation is that market timing DOES NOT work on a consistent basis. Given that fact, not knowing when the market will go bear or will start back up, staying the course is the only logical choice.

The market timers are offering a false choice, loose money or get out and avoid the bear. The problem is the latter, that it is POSSIBLE to avoid the downs and still benefit from the ups. Understanding the reality of the process makes it easier to suffer the misery of the down times. Asset allocation, considering the investment time frame and investor comfort, is the solution.

-- posted by oneputt

Top 55. Jun 5, 2001 8:06 AM

» Rande - Seriously

From time to time, despite the contentious atmosphere which seems to swirl around a certain guru, a serious discussion of the pros and cons of market timing vs. long-term strategic asset allocation manages to break out. Unfortunately, it soon gets overwhelmed by the cult of personality connected with the guru du jour.

How about a place where those who are serious in their defense of or objection to market timing can debate without all the tabloid stuff on the entertainment thread. Doesn't mean you can't express a strong opinion or be emotional about it, just leave the guruitis behind and use attributable quotes, articles, verifiable performance data, etc. to make your best case.

______________

Thought we could start things off with an article from the Journal of Investing. This one takes a serious look at Marty Zweig's timing model and is fascinating if for no other reason than for once a "proprietary model" is put under the microscope.

The Journal of Investing, 1996 Winter, Vol. 5, No. 4; Pg. 79

MARKET-TIMING STRATEGIES: CAN YOU GET RICH?

By Roger C. Vergin ( professor of business administration at Penn State University's Great Valley Graduate Center.)

If making money in the stock market were easy, we would all be rich. Yet, some "experts" say that you can substantially outperform the market just by following a few simple rules for electing stocks or timing the market. To demonstrate the power of their rules, they often provide as evidence a look back in time to show how successfully the rules would have worked in the past.

One of these experts is market guru, Martin Zweig. Zweig is a panelist on Louis Rukeyser's Wall Street Week and a frequent contributor to Barron's. He has been rated the "most consistent long-term performer" by the Hulbert Financial Digest. He also manages over $ 6 billion in pension funds, mutual funds, and partnerships.

In 1986, Zweig published Winning on Wall Street, or WOWS. In it, he presents our "indicators" for stock market timing combined into a "Monetary Model" and a "Super Model." He modestly labels the latter as "the only investment model you will ever need."

Zweig illustrates his market-timing indicators and models (which we will call Zweig Strategies) using data for periods ranging from nineteen to thirty-three years and ending in early 1985. The models all substantially outperform a buy-and-hold strategy, with annual rates of return as much as eight times as large, according to some measures.

The Zweig Strategies are based on factors many investors believe to be important such as changes in interest rates, discount rates, consumer credit patterns, and market momentum. Thus, they have a foundation in ideas generally accepted in the investment community.

Rules for stock market investment are notorious for showing profits when tested on the same sample period data from which they are developed, then failing when applied to a new period. There is danger in using the same data to both discover and test the rules.

By examining data in enough different ways, one can always find investment rules that would have worked well in the past. In fact, they might represent no more than temporary trends or statistical aberrations. The test of a market strategy must be how well it does after it has been discovered, not before.

The ten-plus years of stock market activity since WOWS was published provide material for an out-of-sample test for the Zweig Strategies. How well do they work? Annual returns eight times better than market averages? Hardly. They do not outperform the averages at all.

In fact, they produce results that trail the market averages by amounts that are statistically significant. Average annual returns are 9.0% for the Zweig Strategies compared to 14.4% for a buy-and-hold policy.

ZWEIG STRATEGIES

The six Zweig Strategies for market timing are:

1. Prime rate indicator.
2. Fed indicator.
3. Installment debt indicator.
4. 4% indicator.
5. Monetary model
6. Super model.

Complete explanations appear in WOWS; we describe the indicators and models briefly here.

1. The prime rate indicator signals buy for any initial cut in the prime rate (PR) when PR < 8%, or on the second of two cuts or a cut of one full percentage point, when PR > 8%. It signals sell on any hike when PR > 8%, or on the second of two hikes, or a hike of one full percentage point when PR < 8%.

2. The Fed indicator uses the Federal Reserve controlled discount rate and reserve requirement. A signal for going in or out of the market is developed based on changes in the rates; decreases are bullish, increases bearish.

3. The installment debt indicator gives a buy signal when the year-to-year increase in consumer installment debt drops below 9% and a sell signal when it goes above 9%.

4. The monetary model uses the sum of the scores from the three indicators above for buy and sell signals.

5. The 4% indicator is a measure of market momentum and gives a buy signal when the weekly close of the Value Line Index increases 4 percentage points or more from any weekly close since the previous sell signal. A sell signal is given when the index drops 4 percentage points from any weekly close since the previous buy signal.

6. Zweig's super model combines the 4% indicator with the three indicators from the monetary model to figure out buy and sell signals.

To show their worth for stock market timing, Zweig compares the returns that would have been obtained following each strategy to the returns that would have been obtained following a simple buy-and-hold strategy.

ZWEIG TEST

Zweig compares five indicators against both the Standard & Poor's 500 composite index (S&P 500) and his own creation, the Zweig unweighted price index (ZUPI) of all the stocks on the New York Stock Exchange (NYSE). The sixth, the 4% indicator, is tested against the Value Line Index (VLI). Thus, Zweig has eleven comparisons.

Meaningful comparisons of the individual Zweig Strategies with the buy-and-hold strategy and with each other are difficult because Zweig changes his evaluation criteria from one strategy to the next in WOWS. Times vary from nineteen to thirty-three years. Three of the comparisons include dividends; eight do not. Some earn interest at 6% when out of the market, one earns 7%, and one uses a variable rate.

Seven of the comparisons are by annual rates of return, while four show rates of return only for the times while in the market. For five of the Zweig Strategies, the policy is simply to sell the stock and invest the cash in short-term instruments when out of the market; the 4% indicator calls for shorting the market during those out-of-market periods.

Despite all the inconsistencies, Zweig's analysis clearly suggests that following any of his market-timing models would have resulted in a substantial improvement over buy-and-hold. For example, for the nineteen-year period, the super model would have produced an 18.0% annual return compared to 6.1% for buy-and-hold when measured against the ZUPI. The difference seems even more dramatic in dollar terms. An initial $10,000 investment would have grown to only $30,713 under buy-and-hold compared to $237,509 using the Zweig super model (WOWS, p. 114).

POST-PUBLICATION TESTS AND RESULTS

To make a post-publication test meaningful, I reevaluate the six Zweig Strategies using consistent criteria for his original period, and using the dates for going in and out of the market from WOWS. For all eleven comparisons, both the Zweig Strategies and buy-and-hold are evaluated using a dividend rate of 3.5% per year when in the market and an interest rate of 6% per year when out of the market. Both numbers are close to the average values over the time. Results are in Exhibit 1.

As before, the 4% indicator is evaluated assuming the market is shorted when the investor is not in a long position. Since the total time when long in the market is about equal to the time when short, it is assumed that dividends earned when long are equal to dividends paid out when short.

Using the consistent evaluation criteria, the Zweig Strategies still clearly outperform buy-and-hold. All eleven comparisons show superiority, with an average rate of return of 14.2% for the Zweig Strategies compared to 8.6% for buy-and-hold. To confirm the superior performance of the Zweig Strategies, a t-test for paired data is run on the rates of return. The difference is significant at the 0.0001 level with a t-value of 5.98.

An initial $10,000 would have grown to an average of $124,396 under the buy-and-hold strategies; it would have increased to an average of $418,375 using the Zweig Strategies.

The adjusted comparisons all show impressive returns as a result of using the Zweig Strategies. The differences are dramatic and of sufficient size to attract the attention of the investment community. Academic economists might view the Zweig analysis as evidence contrary to the fundamental insight of the efficient market hypothesis that stock prices reflect optimal use of all available information.

A test of the Zweig Strategies over an independent time fails to support that conclusion, however. I tested the six Zweig Strategies for the period beginning with the last buy or sell signal published for each in WOWS through December 5, 1995. The S&P 500 is again used. Since Zweig's personal index, the ZUPI, is not publicly available, the Value Line Index is used as a close substitute. (VLI is an unweighted index of approximately 1,700 stocks, most of which are on the NYSE; ZUPI is an unweighted index of all the stocks on the NYSE.) The Dow Jones Industrial Average is also added as a base for comparison.

To decide when to move into and out of the market, I track the signals Zweig uses (such as changes in the prime rate, or consumer installment debt that have occurred since 1985), and use Zweig's numerical scores for buy and sell decisions. The resulting movements into and out of the market, with the values of all three indexes, are in Exhibit 2.

Rates of return are computed for each of the six Zweig Strategies and compared to buy-and-hold against all three indexes. As before, a 3.5% dividend rate and 6% interest rate are used. For the 4% indicator, the assumption is again that the market is shorted when not in a long position. Transaction costs and income taxes are ignored; thus, a positive bias remains for the Zweig Strategies, with their more frequent transactions.

Using the times in Exhibit 2, the returns of the six Zweig Strategies compared to buy-and-hold, measured against all three indexes, are shown in Exhibit 3.

The results show that, over the ten-year-plus period, none of the six Zweig Strategies outperforms a simple buy-and-hold strategy when compared against any of the three indexes. For a portfolio divided into six equal parts and each part managed using one of Zweig's six indicators or models, the average annual return would have been 9.1%, compared to an average return of 16.2% by simply following buy-and-hold, compared against the S&P 500. When compared using the Value Line Index, the average annual return for the six Zweig Strategies is 7.7% compared to 9.3% for buy-and-hold. When compared using the Dow Jones Industrial Average, the average annual return is 10.1% for the Zweig Strategies compared to 17.6% for buy-and-hold.

In the eighteen comparisons, the Zweig Strategies produce inferior results in all eighteen cases with an average 9.0% return compared to 14.4% for buy-and-hold. Exhibit 4 summarizes the Zweig Strategies compared to
buy-and-hold for the independent time test (Exhibit 3) compared to the values from Zweig's pre-publication analysis (Exhibit 1).

To find out if there is a significant difference in the performance of the Zweig Strategies compared to buy-and-hold, a t-test for paired data is run on the rates of return. Buy-and-hold is superior by an average of 5.4 percentage points. The difference is significant at the 0.0001 level with a t-value of 4.63.

CONCLUSIONS

The analysis shows that the Zweig Strategies fall into the same category as many other stock market-timing models. That is, although they appear very impressive when illustrated against a past time, they fail to produce superior returns in an independent test. Indeed, for the past ten-plus-year period, they produce results significantly inferior to the simple strategy of buying stocks and holding them. This supports the conclusion that Burton Malkiel draws from a comprehensive examination of market timing research in A Random Walk Down Wall Street that "market timing is likely, not only to not add value . . . but to be counterproductive (emphasis in the original) [1990, p. 178]."

While it would be nice to have comparisons over a longer time, the ten-plus-year period is of sufficient length to raise serious doubt that the Zweig market-timing rules have useful value. By trying enough patterns against past events, one can always find simple rules that would have worked well in the past -- "data snooping" is the term used to describe this process. My research shows the danger of following such arbitrary rules, even when it appears they would have worked quite well in retrospect.

Zweig's indicators are based on ideas for moving into and out of the market that are widely accepted by knowledgeable market observers, and many would suggest it is worth monitoring factors such as changes in the prime rate, the discount rate, and the consumer debt pattern. To base an investment policy on a particular numerical value that would have worked well in the past, rather than on a more comprehensive evaluation of the economy and the market, however, can prove highly unsatisfactory.

Suppose investors had used Zweig's 4% indicator. They would have been whipsawed into and out of the market twenty-one times in the past ten years, and earned an annual return of only 2.8%, compared to the S&P 500 index. The investor who just bought and held would have earned a 16.3% annual return.

The conclusion that must be drawn is that Zweig does not have "the only investment model you will ever need."

-- posted by Rande

Top 56. Jun 5, 2001 9:09 AM

» Rande - Re: Seriously

In response to message posted by Rande:

Forgot we had this thread. Guess there's not much call for serious debate on this issue, after all.

-- posted by Rande

Top 57. Jun 5, 2001 9:13 AM

» KLR - "there's nothing to time [in a bull market]." -Martin Zweig

In response to message posted by Rande:

I guess Marty forgot to include a deflation indicator...but newer models may well include that..."We might have to go with some other fail-safe system."

Kiplinger's Magazine
April 1999

- Mutual Fund Hall of Shame - page 9
Zweig Strategy Fund A
For years market timer Martin Zweig told anyone who would listen that the reason this fund was badly trailing the market was because "there's nothing to time." He's right. When the stock market goes straight up, you can't expect a market-timing fund to keep pace.

The bear market from July 17 through October 8 offered an opportunity for Zweig to strut his stuff. Instead, he fell flat on his face. While the S&P 500 lost less than 20% during that three-month downturn, Zweig Strategy Fund A, which has a maximum sales charge of 5.5%, plunged 25.8%.

David Katzen, 41, who selects the fund's stocks and manages the fund using Zweig's timing signals, says Zweig's indicators remained mostly positive throughout the sell-off. The market was headed down, which is bearish. But bond yields were falling and many investors were pessimistic--both of which are bullish signs to Zweig.

The fund has held an average of 12% of assets in cash over the past three years. At one point in 1994, it had as much as 65% in cash. It entered the 1998 bear market with about 85% of assets in stocks, which it trimmed to 72% near the end of the decline.

The problem, says Katzen, is that Zweig's indicators are designed to protect against the kinds of bear markets that have been prevalent over the past 50 years--those that are triggered by rising inflation and interest rates. But it was fear of deflation that sent the stock market down last summer. "We always said we might have difficulty with deflation," he says. "We might have to go with some other fail-safe system." Another problem, beyond poor market timing: The undervalued stocks in the fund were no match for the growth stocks that propelled the S&P 500.

Whatever the reason, the bottom line is clear: Zweig Strategy, launched in 1989, is in the bottom 10% of long-term-growth funds over the past three and five years. Martin Zweig has agreed to sell the funds, and he could relinquish his market-timing responsibilities in as little as three years.

[Other losers in the Hall of Shame]

Page 1: Introduction
Page 2: Bull & Bear Special Equities
Page 3: Dreyfus Aggressive Growth
Page 4: Fidelity Trend
Page 5: Frontier Equity
Page 6: Lindner Growth
Page 7: Perkins Opportunity
Page 8: Robertson Stephens Contrarian A
Page 9: Zweig Strategy Fund A
Page 10: The Hall of Shame

-- posted by KLR

Top 58. Jun 5, 2001 9:16 AM

» Kirk - Re: Seriously

In response to message posted by Rande:

Nice article Rande.

As an engineer with tons of training in modeling and many decades of application, I always get a kick when I read stuff like this:

Study Fourier Transforms [ http://aurora.phys.utk.edu/~forrest/pape... ]and you will see that this statement was proven by a French Mathmatician long, long ago. ANY event can be modeled as a superposition of sinusoids of different frequency which sum to the original waveform.

You can actually make this give waveforms that look like square waves!

The problem is once you go outside the period you set your model for. You can model ANY PAST EVENT with near 100% accuracy with a Fourier transform but it will often fail miserably one second after of the last data point taken (i.e. it will NOT predict the future.)

By fail, I mean the model's output often literally blows up! It is really fun to do on a computer and plot it just to see how poorly it predicts the future! (BTW, this technique has been used for decades to design filters where the filter works very well over the period covered but blows up outside the band (1/time) of interest.

Nothing that Marty or Brinker can or could do would be any better than what a BIG computer can do. The trick is they can assign all the economic, sentiment, etc indicators to the major Fourier components and TRY to predict the future, but "real life" is not a "continious event" then ALL the models eventually HAVE TO FAIL as that is where the Fourier Transform fails.

My most recent experience with modeling is in predicting light absorption and emission of compound semiconductor devices (GaAs or InP based rather than Silicon based) and these have problems also due to quantum physics which says (man am I simplifying here) that optical events are discrete events that you model with probabilities... You have to find enough "linear events" to model to make any sense out of what is happening... but I digress...

Simply put, I doubt guys with finance degrees are doing Fourier modeling, especially as far back as 1988 where I only know of a few that were doing it for optical transmission down a fiber (I worked with the guy doing this... and he is now a senior scientist at Agilent - VERY highly rated guy...) and even if they were, it probably won't work to predict the future!

smile

-- posted by Kirk