Mutual Fund Timing: Lessons From The Past

The last decade has been essentially flat in terms of stock market return. The timing of mutual funds has decreased substantially following the documentation of frequent trading as well as illegal late trading. Many large mutual funds have severely clamped down on moving dollars between funds, essentially locking an investor into a long-term buy & hold mode.

There are many advocates of this approach. However, those investors who adopted it and fully invested in the stock market around March 2000 took a beating, with losses often exceeding 60% to 70%. It has been in the interests of many mutual funds to limit exchanges between funds, since it contains costs. But is it in the best interest of investors? Should investors be locked into very long holding periods, given that many large funds such as ING, holding billions of dollars in retirement assets, were on the verge of going belly-up prior to the government bailout?

The fundamental tenet of any timing technique is to find a repeatable opportunity in terms of entry and exit that allows a reasonable chance of profit while limiting market exposure and then exploit this to provide an acceptable return with reduced risk.

Perhaps the best example of a successful mutual fund timer was fund manager Gil Blake. Of his "futures-type" market returns -- about 45% with a minimal drawdown, for example -- 3% has been well documented by analyst and author Jack Schwager. Similar mutual fund timing exploits by investor Steve Lescarbeau have also been documented by Schwager.

METHODS
While we have no personal knowledge of the exact systems used by successful mutual fund timers, we can fall back on tried & tested strategies that they may have adopted for success. A basic well-known market timing strategy is to buy on strength and sell on weakness. Let's see if we can codify this into a system that can be tested on a mutual fund.

Given the upward bias of the stock market, this would be a long-only system. In general, we want to choose a mutual fund that has built in volatility to take advantages of prospective moves. We used Invesco Technology (FTCHX) to develop our model and then test it on another high-beta fund, T. Rowe Price Science and Technology (PRSCX). The mutual fund data was downloaded from Yahoo.

RULES
The rules are: buy when the fund's close is expected to be positive compared to the previous day and sell when the fund's close is expected to be negative compared to the prior day.

The following code was developed for the Behold testing platform:

 If close > close[1] then buy today at close;
If close < close[1] then exitlong today at close 

There would be a certain amount of estimation involved since the trade would have to be placed before 4 pm Eastern time. A fair estimation of the closing price could be made at about 3:55 pm, either by looking at the fund's largest holdings or a major correlated index. The model returns were tested on a percent basis.

FIGURE 1: TEST RESULTS FOR FTCHX (1986-99). There were around 700 trades with about 49% profitable, a profit factor of 2.6, and a maximum drawdown of about 18%.

However, take a look at the result from Figure 2, which indicates its performance has drastically diminished 37% profitability, profit factor of 1.2, and a drawdown that has increased to 66%.

FIGURE 2: TEST RESULTS FOR FTCHX (1999 to 2/12/2012). The performance dwindled to 37% profitability, a profit factor of 1.2, and a drawdown of 66%.

The combined equity curve for FTCHX for both periods also highlights the break in the linear uptrend starting around 2001, as shown in Figure 3. The equity curve is very similar for PRSCX (Figure 4).

FIGURE 3: EQUITY CURVE FOR FTCHX (1986 to 2/12/2012). Note that in 2001, there was a break in the linear uptrend.

FIGURE 4: EQUITY CURVE FOR PRSCX (1986 to 2/12/2012). The equity curve is similar to that of FTCHX.

CONCLUSION
Several important points can be gleaned from these results. First, no model will work forever. Not only is the market changing constantly, it is not immune to inappropriate regulation, poor political decisions, and so on. If there was a concern about market timing hurting long-term investors, then a nominal fee such as 0.1% of the transaction amount could have been charged. Many funds imposed charges that were impossibly high, such as 1% to 2%, and this essentially made timing mutual funds impossible. Ironically, the funds themselves may have done the long-term buy & hold investors more harm than any timing endeavor, with FTCHX declining over 80% and PRSCX declining about 65% in the approximate two years following April 2000.

What caused the model to regress and become nonviable for trading these two funds? There are likely several reasons. The practice was becoming more prevalent and some large funds were indulging in this practice as well as doing illegal after-hours trading. More and more mutual funds started putting barriers to mutual fund timing that were increasingly difficult to overcome.

While not everyone will agree with the concept of timing mutual funds, we can all learn from a system that was elegant and highly profitable for over 13 years.

SUGGESTED READING

• Schwager, Jack [2001]. Stock Market Wizards, HarperBusiness.
• _____ [1992]. The New Market Wizards, HarperBusiness.

This article is dedicated to Gil Blake and Steve Lescarbeau for reminding the trading world that a simple robust approach is the key to profits.