  |
 |
FRED
GOODMAN'S KEY INDICATORS FOR INVESTMENT SUCCESS
The Daily Rating can be used as an aid in selecting the day to enter or exit a position that one has already decided to trade, based on unrelated analysis. It is hoped that by buying on a day expected to be weak one can get a better price than at other times -- you don't have to be a day-trader to profit from the edge that the indicator provides. The Daily Rating is based on variations in market performance that have been found to fall on different days of the week and days of the month. A day is considered to be a good day to sell when the S&P 500 is expected to close higher, and is expected to be weak for the following three days. A good day to buy is one on which the S&P is expected to close lower if the next three days are expected to be strong. It is important to keep in mind that this is just one indicator, it must be used in conjunction with others. For example, if the Daily Rating is negative while the Average Signature is saying buy, then obviously one must take both into account when deciding on the size of the investment and its timing. I have evaluated the system going back to mid-1996. Starting on July 18, 1996, if one had taken long or short positions in the S&P 500 based solely on the Daily Rating, $1000 would have grown to $4196 at the close on October 9, 2006 (pink line). There were good times and bad as you can see by the 48.6% drawdown from the peak of $3720 reached on July 29, 2002, and the low that occurred on August 6, 2003. A drawdown that big would have shaken out all but the most hardy, but it is based on the raw system, without considering any other indicators that might have been employed.
Even with that big drawdown the system produced a 14.6% compounded annual return over the ten and a half year period. At the same time, being fully invested in the S&P 500 generated only 7.9% compounded, and it suffered a drawdown of 87.1% between March 27, 2000 and the July 29, 2002 low. So all-in-all the result is worth pursuing, and we will do so. In addition to the buy/sell strategy, the chart above shows the buy-only (blue line) and short -only (green line) results as well. While they produced less than the combined strategy, each must be credited with the prevailing interest return for the time that the cash was not invested. For the buy-only strategy this amounted to roughly 3% interest for 3.7 years and resulted in an overall compounded annual return of 11.6%. The short-only strategy produced a compounded annual return of 3.8% including interest. An attraction of the buy-only system is that it suffered only an 18% drawdown, and it has acheived a new high, whereas the S&P 500 had an 87% decline and has not yet regained its 2000 high -- on a risk-adjusted basis there is no comparison between the buy-only and a buy and hold program. The Daily Rating is determined each day by looking at a one-year moving average for each day of the week and day of the month, and then rating each of the next three days. Let's take a closer look at the method behind this indicator. The next chart shows the one-year moving average of the daily return of the S&P 500 looking only at Mondays, since July 1996. The percent return (pink line) is plotted against the S&P. Based on 11 years of data, Mondays have produced the best return of any weekday. However, it is clear that there have been considerable variations in return during the period.
Here is Wednesday plotted on the same scale. It is considerably more volatile than Monday, and while the 11-year return is not as good, it is currently beating Monday's average return, 0.19% to 0.02%. The Daily Rating takes these variations into account in assessing the expected strength or weakness three days ahead.
This next chart shows the one-year average percent return for the first day of the month. It has been the strongest day during the 11 year period and it is the most consistent day as well.
The variation on the other 30 days of the month is greater than that found for the first day, but all of the variations are considered in formulating the Daily Rating.
Historical Background In the course of the last few years I have covered timing aspects related to the day, week, month and presidential election cycles. In this section I have assembled much of that material to provide background for the Daily Rating. In the next two charts we'll look at fifty-years of presidential election cycles -- we last looked at them three years ago. The two year periods that end with a mid-term election -- like the one we will have on November 7 -- are in white. Since 1985 the market has moved higher in every white period other than the bear market that started in 2000, but even in that difficult period there was a rally associated with the election that was held at its end. Notice also that there was always a dip near the end of each white section, just as we had in May. I have been anticipating weakness as we approached the election, but so far it has not materialized.
The period that started in 1958 was much more violent than the steadily advancing market we have enjoyed since 1982. True the log scale does work to improve the appearance of the turmoil that existed between 2000 and 2003, and also the sharp drops in 1987 and 1991, but compared to the years charted below, it was a romp. Even with all the violence between 1958 and 1981 the blue periods were relatively placid compared to the white, and we are about to enter a blue period in November. Of course it is possible that we have entered into a secular bear market after the long bull market between 1982 and 2000, but it is also possible that we have simply returned to the days of steady measured growth that occurred between 1984 and 1995 -- a quiet advance interrupted by occasional downdrafts. If that is the case, then the 5-year bubble and its 3-year reversion to the mean, may simply have been a violent storm in the middle of a long period of economic growth -- fueled by a tremendous increase in productivity -- that has not yet ended.
Let's break it down further, and look at what historically have been "good months" and "bad months" for stocks. The "good" months are January through April, July, November and December. The "bad" months are May and September, and June, August and October are middle of the road. In the next chart I have superimposed the good and bad months upon the concept of good and bad years in the presidential election cycle since 1995, and have included the S&P 500 for comparison. I think you will see that the oscillations of the election cycle and the good and bad months have coincided reasonably closely to the S&P. It is certainly not perfect, but if you look very closely you will note that the major declines and advances seem to have come pretty close to the lowest and highest points in the pink line.
Let's drill down even further. I have often called attention to the finding that the first five days of the month are the best five days. However, when I extended the original investigation that covered the period between 1995 and July 2003 to the present, I found that there have been changes in the last three years. As you can see in the next chart, the very best day since July 2003 was the 28th of the month. This is based on the assumption that $1000 was invested in the S&P 500 (or SPY) at the close of the trading on the day before the 28th, and sold five days later. Second best was the 26th, and third place was a tie between the 1st and 25th. Are these important differences? I think not, but read on to see what you think.
This chart was taken from my December 3, 2003 report, and you can see that buying at the close of the day before the 1st provided the best 5-day return during the first 8-1/2 years, by a considerable margin. The next chart shows how the inclusion of the recent three years has affected the results.
Really very little has changed. There is still an edge if one buys towards the end of the month, but for long term investors (anything longer than a few months), buying near the 12th or 28th, and selling near the 6th or 16th is statistically the best plan. This is true since you want to buy just before an increase and sell just after it.
Here I have added the blue line which represents the net positive closes for each day of the month covering the 11 year period. There is no question that both the net return from a 5-day investment starting on the various days of the month (pink line), and the net number of positive closes are well correlated. On average, a 5-day investment rose 56% of the time over the 11 year period. However, when the investment was at the close of the day before the first or before the 30th, gains were made 70% of the time. Buying on the days between the 4th and the 8th, or those between the 17th and the 24th resulted in 5-day losses more frequently than gains. Consequently, if you are considering a purchase of an equity that you plan to hold for a period of months, you will have an edge if you buy it just after those losing days (around the 12th and 28th), and when you sell you will have an edge if you sell on a day when the odds are that the market is likely to fall for the next 5-days, such as the 6th or the 16th.
Here are the results for 1-day investments. The days with the highest frequency of closing down stand out, they are 7-10, 19-24 and the 30th. Those are the days on which we should try to buy into the market whenever possible. Sales are best planned for the days that most often close higher.
The final two charts today compare an investment opened on the final trading day of the month and sold 5 days later, to one held for a single day. There is very little difference between the 5-day and 1 day holding periods when the final day of the month is chosen for entry. Of course the return was higher for the 5-day investment, but money was tied up for 5 times as long. Again, I am not suggesting that we all become day-traders, this information is to help us choose sound entries and exits regardless of the length of the holding period.
Next week we'll add the final leg to the stool when we discuss in detail which days of the week have proven to be best for entry and exit, and I'll try to tie it all together in a strategy that can be monitored daily as changes occur.
As a teaser, the first chart summarizes the differences between the days of the week, and it is clear that Monday was a winner while Thursday and Friday brought up the rear. However, there have been clear changes in the course of the 11 years. For example, notice that Monday (blue line) has been good in bull markets and poor in bears. A very important factor separating success from failure in a trade is slippage -- how much a stock ticks down when it is sold and how much it ticks up when it is bought. The best way to minimize slippage is to sell when the market is rising, and buy when it is falling. Since Mondays have been the most reliable gainers, it is the choice day to sell stocks, and since Thursday and Friday are typically losers, they are good days to buy. More on the subject next week, but let's take note of the fact that Fridays are not only one of the two best days to buy, today is also the last day of the month, which has also been a winner. It suggests that even if the Dow fails to make its new high today, there is an excellent chance that it will make it on Monday.
I have been working on an investment calendar to quantify the edge that is available when one buys on traditionally weak days and sells on strong ones. The good news is that today and tomorrow are the best days of the week to buy, but the best buying days of the month are between 6 &13 and 20 & 25, so this week, Friday is the best day for buying. Both the day of the week and the day of the month have been found to offer significant and repeating advantages. There is a clear-cut advantage when one sells on a Monday since Mondays are generally strong, and I have found that the advantage decreases daily until Friday, which has provided the smallest average return of any day of the week during the last 11 years. However, when you look at the frequency of gains as opposed to the size of the gains, between Monday and Friday there are no differences. Each day of the week since 1995 has closed up between 51 and 53% of the time and the standard deviation was twice that, so there is no significant daily difference in the number of positive closes. However, the size of the gains varies by a factor of two between Monday and Friday, with Monday providing the biggest average gain and Friday providing the smallest. Monday and Tuesday are pretty close to each other, as are Thursday and Friday. No question about it, if you're planning to sell, you'll have an edge if you do so on Monday or Tuesday near the close. And if you're planning to buy, Thursday and Friday are the best days to do so. However, it is interesting that of the ten largest one-day declines, five of them occurred on a Monday, and in spite of that, Monday still has the highest average return. The other five big declines were divided between Tuesday, Thursday and Friday. Wednesday is a follower -- it goes with the trend and has a middle of the road return, but it had the smallest maximum loss and the biggest maximum gain of any day of the week. Nine of the ten biggest one-day advances in the last 11 years were spread evenly with three on Tuesday, Wednesday and Thursday -- the tenth big advance fell on a Monday. Here I have collected all of the above information together to reflect all three cycles/seasonals -- daily, monthly and election. It serves as an interesting guide to the market, but clearly it does not represent a standalone system for investing -- nothing does. There certainly is a correlation between the chart and the market from 1997 through 2003, but it is not as clear a relationship since then. Perhaps they will relate well in 2007 when the election cycle moves into its strongest phase -- time will tell. That said, I do believe that a slight edge is available from buying and selling on the proper days of the month and will try to be guided by that going forward. The next few charts show some of the key days.
$1000 invested on July 3, 1995, held for 5 days and reinvested on the last day of every subsequent month until August of this year, would have resulted in a total value of $2315.71, for an annual return of 11.7% for 11-1/4 years, not including the interest you'd earn on your idle cash all the days of the month you weren't in the market (probably another 2-3%). This is a remarkable result considering that money was at risk only about 25% of the time! The most severe drawdown was just 7.3% between March 2002 when the investment was worth $1926, and October 2002 when it was worth $1786. The results during the last three years were not as good as they were for the total period. $1000 invested on August 1, 2003, and held for 5 days from the last day of each month resulted in an annual return of 5.6%, and if invested on the day before the 28th day it returned 6.4% for the ensuing 5-day period. Nevertheless, there is an edge, and that's what we're looking for.
Here is the best return day of the last three years it measures the return from buying at the close on the 27th and selling at the close 5 trading days later. While it was best for the last three years, for the full 11 years the return was just $1567.88, or 5.1% annually. The biggest drawdown was 15.1% which is more than double that resulting from buying on the final day of the month. This illustrates that there is variation and that these guidelines should be used flexibly for buying and selling rather than as a rigid "system." Furthermore, I have not taken into account the commissions and slippage resulting from the frequent trading. Clearly the benefit of selecting the day comes from timing longer trades, not from constantly moving in and out of the market.
Here is a peculiar day -- the 3rd -- in that it dramatically changed when the market switched from bull to bear and has remained near the bottom ever since.
Buying on the first day of the month rather than at the close of the last day of the preceding month resulted in just a bit more than 1/2 of the profit and a bigger drawdown.
Finally, here is the day that was consistently the worst -- the 7th. It resulted in a loss of 24% over the 11-1/4 years. Clearly it was better to sell at the close on the 6th and to buy five days later than on most other days of the month.
Fred Goodman, CFP, is a fee-only Certified Financial Planner based in Los Angeles. You can send him your questions and comments via email at Fred@MarketMonograph.com. The charts and commentary represent what Fred is thinking about the market and thinking of doing for his own account and for accounts he manages. There is no guarantee that you will profit from trading as discussed herein. You may lose money and Fred assumes no responsibility for what you do or do not do with this information. 
|
