Safehaven | Making Better Investment Decisions with the Help of Empirical Research (Part II)

August 07, 2008

Making Better Investment Decisions with the Help of Empirical Research (Part II)
by Tom Madell

Summary: In my continuing research on fund investing, I have assembled large amounts of data on mutual fund performance. This data has proven useful in yielding insights on questions of crucial importance regarding how investors might more successfully manage a portfolio of fund investments. I would like to convey these insights to those who are open to applying such empirical research in an effort to maximize the odds of getting above average returns.

In particular, I am trying to validate a straightforward set of rules to help investors make good buy, sell, and hold decisions based on our data which covers performance results going back to early 1995.

---

We have already presented some of our initial findings on our website in our July newsletter. (Click here to read the article and for information on how we set out to define a classification system which could be regularly applied to each of the major fund categories to determine whether that category was currently a BUY, SELL, or HOLD.)

We now can present data that seems to clearly show the merits of using this classification system in helping to make each of these 3 crucial portfolio decisions. In particular, we can now document that using such a system would have resulted in improved overall portfolio results as compared to less systematic approaches, or merely just buying and holding. The results were particularly dramatic over the last 8 1/2 years.

Results

First we show performance outcomes for stock fund categories classified as BUY, SELL, or HOLD for the period between Jan. 2000 and July 2007. The ensuing performance for funds so categorized was examined through the close of June, 2008.

The table, and the two that follow, shows what the average return would have been for many of the major fund categories (such as Large Growth, Small Blend, International, etc.) that we classified using the rating system we developed; the figures thus show what the typical return would have been 1, 3, and 5 years later, assuming one had bought, sold, or held a fund in that category at the time the signal was first generated. (The 3 and 5 year returns are annualized, that is, they are average per year returns.)

The returns shown for funds categorized as SELL are the returns the investor would have received had he/she continued to own the fund without selling. Returns highlighted in red indicate performance that we consider deficient and therefore problematic for investors.

Recent Data for Fund Categories Classified as BUY, SELL, or HOLD
	1 Yr Return	3 Yr Return	5 Yr Return
BUY	11.9%	11.9%	15.0%
SELL	- 11.2	- 4.3	2.8
HOLD	10.7	12.0	12.2

Note: Three and 5 yr returns beginning July 2005 and July 2003 respectively.

The next table shows data for classifications BUY, SELL, or HOLD made for the period between Apr. 1995 and Oct 1999. The ensuing performance for funds so categorized were examined through the close of Sept, 2004, that is, up to 5 years after.

Earlier Data for Fund Categories Classified as BUY, SELL, or HOLD
	1 Yr Return	3 Yr Return	5 Yr Return
BUY	NA	NA	NA
SELL	23.9	8.2	5.0
HOLD	22.5	14.5	9.0

Note: There were no BUY classifications made during this period.

The last table shows the results for all BUY, SELL, and HOLD classifications made using the combined data over the entire period studied, namely between Apr. 1995 and July 2007 with performance results through the close of June, 2008. That is, it incorporates all data together from both the first two tables.

Combined Data for Fund Categories Classified as BUY, SELL, or HOLD
	1 Yr Return	3 Yr Return	5 Yr Return
BUY	11.9	11.9	15.0
SELL	6.6	2.2	4.0
HOLD	13.1	12.7	11.1

Note: No BUY classifications were made prior to Apr. 2003.

The Important Implications for Investors

It is important to consider these results carefully as they give important support for the validity of each of the BUY, SELL, and HOLD classifications of stock fund categories that we empirically created.

Had there been no discernible patterns in prior stock category performance, it would have likely been extremely difficult, if not impossible, to create a rating system that would work effectively for a period as long as 8 1/2 years, the initial study period of used in setting up the criteria for our classifications. However, rather than using a rating system based totally on the observable results over the 8 1/2 years, we relied heavily on our previous research which indicated that fund categories that appear overvalued and/or showing negative but sustained recent performance, consistently were less likely to do well in upcoming years as compared to funds showing the opposite characteristics.

In applying our classifications to the earlier 4 1/2 year period between 1995 and 1999 and examining the results, we were able to substantially confirm the usefulness of our ratings using independent data.

In particular, our results suggest the following:

Using our SELL signal helps an investor avoid, at worst, potentially negative or sub- par returns for a fund category. During the last 8 1/2 years, had we had use of our SELL classifications for major fund categories, we would have been able to steer clear of funds that suffered the most, especially during periods where the markets showed major downturns.

Note though that during the 1995-1999 period, our SELL signals initially proved to be premature. That is, for the 1 year period following a SELL warning, fund categories so classified actually did slightly better than other fund categories. But subsequent to that, a SELL fund category's performance did indeed show significant deterioration, pulling one's annualized return on the average fund to more than 6% per year less than for other fund categories (and 4% per year less even after 5 years had elapsed).

The reason for the difference between the 1995-1999 and 2000-2007 sub-samples is clearly due to the fact that during 1995-1999, we were in a non-stop bull market. Thus, even if a fund category was overvalued, it tended to keep going strong anyway. But by 3 and 5 years later, our SELL-designated categories had lost nearly all of their initial 1 year gains, thus showing virtually no annualized gains the following few years. (For example, a fund that gains 24% for the first year, but then drops to 8% annualized after 3 years, has still gained a total of just 24% - 8 x 3 - or no gain as measured in years 2 and 3.) Thus, regardless of when you encountered a SELL signal, you would have eventually wound up considerably better off elsewhere than in a SELL fund category.
Using our HOLD and BUY signals will likely lead to the best results for investors who want to remain invested in the stock market for periods up to at least 5 years.

When BUY signals are issued, they should indeed lead to the best performance as compared to the other two classifications, and our data does suggest that this is true for our more recent data, especially over the following 5 years even though our 15.0% 5 yr. result is based on just a few signals.

It is important to point out that the BUY recommendations generated by our a priori model of category performance are, by nature, conservative and will only appear when a category becomes quite undervalued and/or the category shows very positive signs of on-going strength. In fact, our classification system issued no long-term BUY signals at all during the 1995-1999 bull market phase all the way until Apr. 2003 and beyond when a select and then growing number of fund categories finally began showing the required performance characteristics needed to qualify as safe BUY signals. You might consider our BUY signals as "super -BUYS" indicating an unusual opportunity that comes up infrequently to do quite well.

For practical purposes, then, since outright BUY signals may be few and far between, fund categories designated as HOLD may be the only place to be if one wants to identify the "best available" current categories (unless we are able to modify our classification system to allow for more BUYS). In fact, whether you actually hold on to an existing fund you already have, decide to purchase more of that fund, or actually start a new position, a HOLD classification means just that - it is suitable to hold (ie own) that fund.

What Does Our Category Classification Recommend Investors Do Now?

At the present time, unfortunately, all the major fund categories would be classified as SELL, and the majority of them have correctly (thus far, showing double-digit negative performance) been SELLS since last Oct. This appears similar to the first two quarters of 2000 when nearly all categories were also SELL. (The performance results after those signals were indeed poor with just a few exceptions.)

However, it does appear that if Large Cap funds (Growth, Blend, and Value) drop approximately 5% to 7.5% below their current levels, they will be reclassified as HOLDS. This could easily occur within the next several months, so it might not make sense to SELL from these holdings at this point. (Our Model Portfolio is about 90% invested in Large Cap funds.)

If the bear market continues to grind on, as we currently expect, some outright BUY opportunities should indeed eventually appear, as first happened at the end of the 2000-2002 bear market. At that time, our system would have identified both International stock funds and Large Growth funds as places to be.

Incidentally, you might wonder if I will publish the criteria I use to determine my BUY, SELL, or HOLD classifications. I offer some suggestions as to how I do this in my July '08 newsletter. I don't plan to publish them in their entirety in the foreseeable future because they may change after further analysis. However, I do plan to use them as the basis for my quarterly Model Portfolio selections going forward. Thus, you can allow yourself to achieve the benefit of their usefulness without having to do any work yourself! Just continue to read our new Newsletters at my http://funds-newsletter.com website.

Tom Madell, Ph.D.
Publisher
Mutual Fund Research Newsletter

Mutual Fund Research Newsletter is a free newsletter which began publication in 1999. It has become one of the most popular mutual fund newsletters on the Internet, ranking number 7 on the Google Directory page for Mutual Funds News and Media Newsletter websites. Since we began publishing our quarterly Model Portfolios, 29 out of the 36 Stock Portfolios have outperformed the S&P 500 Index over the following year (data as of Sept., 2009). Since 2000, 28 of our Model Stock Portfolios have had the opportunity for at least 3 years to pass since we first published it. 25 out of 28 of these Model Stock Portfolios have outperformed the S&P 500 Index over the following 3 years, and ALL Stock Portfolios (20 out of 20) have outperformed over 5 years as well. Mean level of outperformance has ranged from about 3% over 1 year to over 4% PER YEAR over 5 years.