This article will explore one simple technique to improve market returns and reduce risk. The simplicity and improvement are likely to surprise you. 

Market Returns Thus Far

We looked at market returns without diversification and concluded that while the returns were attractive the risk was not. 

This diversification reduced risk to more tolerable levels but reduced returns to where most would consider them unattractive.

The old wisdom that an investor “can eat well or sleep well” has thus far held. Attempts to master the return-risk problem have been somewhat superficial and not very productive thus far. Now we begin the exciting part of this series where we investigate whether eating well and sleeping are truly mutually exclusive.

The tool introduced today may be known to most. It is clearly usable by anyone. Despite these factors, most people do not understand how this tool can produce dramatic improvements.

A Market Filter

A market filter is something that tells an investor when he should be in the market and when he should be out. A filter adds a mechanical aspect to trading. Many investors benefit from the lack of emotion associated with a filter. Often investors get so emotionally invested in markets that their well-formulated plans are abandoned. A filter can make it easier to “stick with the plan”.

Filters can be simple or sophisticated. Numerous technical indicators can be used. Many have been modified to serve as filters. We will look at a simple moving average, the simplest and oldest of technical indicators.

(Click on image to enlarge)

Any average requires a series of numbers. This series is added up and divided by the number of observations to obtain an average. In the case of markets, usually, the closing price is used in the series.

If we add up the last n period closing prices and divide by n, we have created an average. This point is plotted on a chart showing prices. If you do that for all price points on a chart, each date will have a dot representing the average of the last n day prices.

Connect these dots and you have plotted a simple moving average. The illustration to the left reflects daily closing prices and a simple moving average (SMA) based on 200 days.

Characteristics of Moving Averages

There are many types of moving averages — simple, exponential, time-weighted, volume-weighted, etc. For our purposes, only the simple moving average is explored. It is the easiest to calculate and understand. It is also the slowest to react (which many consider a negative and the inspiration for its many forms).

A moving average is a lagging indicator (all indicators are regardless of what is claimed). In the case of a 10-period moving average, it lags because 9 of the numbers in the average are historical and only one is current. In an up-market, prices are increasing and the moving average (eventually) lags behind (it is lower than) price. In a declining trend, a moving average also lags, but it is higher than the current price (older prices are higher than the current price).

The lagging relationship provides the key to using a moving average as a filter or a permissioning switch. If you believe most financial assets tend to trend, then a moving average identifies the trend. If prices are lower than the moving average, then the market is deemed to be in a downtrend. If above, in an uptrend.

Whipsaws, where prices cross above one bar and below on another, can be a problem because they produce a lot of short in and out trades. To avoid the whipsaw effect, some traders require two or more closes above or below before any action is taken. The shorter the timeframe (one-minute bars, 1-day bars, 1 week-bars), the more likely the whipsaws.

In the case at hand, we trade monthly bars. Each trade occurs at the end of the month. Our approach is to accept whipsaws rather than to screen for them. To screen using monthly bars would mean to postpone the decision by one or two months.

Applying A Moving Average

Here is a visual of SPY monthly data and a moving average of 10 (months) for SPY (an ETF for the S&P 500):

(Click on image to enlarge)

The chart shows how the Simple Moving Average (SMA) moves with respect to prices. The trending aspect of the S&P is obvious. To use this simple moving average, you would set a decision rule. Usually, SMA decision rules are set in terms of the slope of the moving average or when a close crosses above or below the SMA.

For testing purposes, the diversified market index VTSMX, a Vanguard mutual fund used in prior analyses, is used to test the efficacy of an SMA market filter. The 10-period monthly simple moving average (SMA) was used. (Note, 10 months on a monthly chart is nearly equivalent to the popular 200-day moving average on a daily chart.)

No diversification was involved except that contained within the VTSMX mutual fund itself. All funds are either in VTSMX or in cash (actually a short-term interest-bearing fund).

The Decision Rule

If the close of VTSMX is above the SMA at the end of a month we stay in the market (or enter if we have been out). Conversely, when a close is below the SMA at the end of a month we go to cash or stay in cash if we were already there.

It is in this respect that the SMA acts as a permissioning switch. The switch is on when closing price > SMA (you are in the market). When the switch is off, you are in cash. The switch is evaluated at the end of every month and action is taken as necessary.

The Results

The effects of using the switch (Timing Portfolio) versus a buy-and-hold strategy based on the 10 mo. SMA are shown below:

(Click on image to enlarge)

The results are spectacular. Using the switch takes the user out of most of the two devastating (50% drawdowns) periods above. The switch produced these improvements:

• Higher total earnings (1.1 million vs. 90o thousand for the buy/hold strategy)
• Higher CAGR ( 10% vs 9%)
• Lower Standard Deviation
• Substantially lower drawdown (18% vs 51%)

To understand why, look at the blue line (the timing portfolio). A level line (with perhaps a short incline or decline) indicates periods where the switch would put you in cash. Eyeballing the above chart suggests the first one was toward the end of 1998 and then only briefly. But look at the major trouble spots for the market (red line) namely the drawdowns of about 50% from late 2000 to mid-2003 and similarly from the beginning of 2008to early 2010. Now, look at the blue line for both periods. Note how it goes level in both periods, avoiding most of these losses.

Had you used this filter, you would have achieved excellent returns without losing a lot of sleep or looking for tall buildings with windows you could jump out of.

A Magic Bullet?

So, have we found a “magic bullet” to protect against bad markets without giving up return? Can life be so simple? The obvious answer is “No,” if only because markets are never easy.

Let’s investigate further by looking at robustness and generality.

Robustness: By “robustness” we refer to the sensitivity of the values used for the SMA. We used 10 months. Do we get similar results for other values? SMAs of eight months and 12 months were checked. Both produced returns in excess of $1.0 million. Drawdowns remained in the 17 – 18% range. The 12-month SMA produced better results than the 10-period SMA, although not meaningful enough to warrant a change. Tests at 6 and 12 months produced profit levels over $900,000 with drawdowns in the 17 – 18% range.

The test appears robust in the sense that it works well for other ranges. (Beware of any test that fails a robustness test. The backtest was likely “fit” to the data to produce the best possible result. Backtesting done this way fails when different data or time periods are used).

Generality: To assess generality, we need to test different time periods and different assets. Several tests are performed below.

Generality Test I — General Electric

General Electric provides one such test. Time-wise, GE is the tale of two companies: early on, it was one of the great corporations in America; more recently a struggling, perhaps failing company. Let’s look at three periods: GE Early, GE Late, and GE Total and the effects of using an SMA screen:

GE Early (1987 – 2000)

(Click on image to enlarge)

GE Late (2001 – January 2019) 

(Click on image to enlarge)

GE Total (1987 – January 2019) 

(Click on image to enlarge)

Aside: You cannot add the two pieces together to come up with the totals for the full period because each period (Early and Late) is started with $100,000.

Observations

An SMA of 10 months was used and applied to GE (not a market index).

GE Early reflects a growth stock similar to Facebook, Apple, Amazon, and Netflix in recent years. The uptrend was so consistent and strong that any crossovers were short-lived and cost you money. Buy and hold beat the SMA strategy in every category but the drawdown. In essence, being out of the market at any time during this growth was not helpful.

GE Late When GE was doing poorly, a buy and hold strategy underperformed the SMA screen on every category. In essence, being in the market at any time during this period was not useful.

GE Total When the entire period (GE Early and GE Late) are looked at as one period, the SMA screen outperformed a buy and hold strategy for every category.

Generality Test II — European Stocks

VEURX, a Vanguard mutual fund of European stocks, was tested with the SMA 10-period screen. The test ran from 1992 to January 2019. The results are below:

European Stocks (1992 – January 2019)

(Click on image to enlarge)

Observations

European results are improved in every category by using the SMA screen.

Summary

The examples above showed the effects of a screen or switch to determine market participation. One should not conclude that such screens work in all circumstances.

Two failures were not shown. By failure, profits were worse with the screen. Drawdowns improved but only marginally. The two failures involved IBM and AAPL (neither shown). No other assets or securities were tested.

Regarding these “failures,” I suspect that the volatility of individual stocks versus the volatility of groups or portfolios of stocks had much to do with the IBM outcome. Individual stocks are more volatile than a portfolio of stocks. In portfolios, some stocks are jumping up because of individual circumstances, others are declining because of their own particulars. Much of the “noise” is thereby nullified.

When a grouping of stocks moves, the move is apt to have more significance than the movement of a single stock. Individual stocks move based on their own uniqueness (surprise earnings, management change, etc.). Such changes can be significant for an individual company but not necessarily create a trend change in a grouping. Signals are also fewer in portfolios, reducing the whipsaw effects.

Stocks like AAPL (or GE Early above) are questionable for screens for two reasons:

1. If they are in their explosive growth period, you probably never want to be out.
2. Even after the explosive growth, the volatility of individual stocks seems to produce too many false signals.

Screening or filtering is an important tool for investors. This article demonstrated why. Consider adding a screen to some or all of your funds.