Market Timing With Ten Month Moving Average: Tactical Asset Allocation Backtest Part 2

What is a ten-month moving average (10 MMA)?

You take the closing price on the 1st of the previous ten months and average them. Of course, there is nothing special about the first and the average could be taken on any day of the month.  You take buy or sell decisions (see below) based on whether the current price is less than or greater than the 10MMA.

Why is an average such as 10MMA used?

Daily market price movements have too many ups and downs. This “noise” makes it difficult to spot price trends. So different types of price averages are used to “smoothen” price movements and spot trends. See the picture below.

How is the 10MMA used?

The 10 MMA is pretty similar to the 200 day moving average. Making investment decisions based on moving averages is known as trend following. If the current price is above the last ten-month moving average, then we expect the future to be bullish and “go all in”. That is 100% equity and 0% fixed income.

If the current price is less than the ten-month moving average then we expect a bearish trend in nurture and exit. That is 0% equity and 100% fixed income. This is summarised in the picture below.

These kinds of investment decisions come under the umbrella of momentum strategies.  That is we look for an upward momentum in price and invest and look for a downward momentum and exit. An upward (downward) movement for the last few days to few months can be expected to continue in the near future too. This assumption is the core idea behind a momentum strategy.

Just like the PE method, the 10 MMA also suffers a similar drawback. The price can move above and below the 10 MMA in quick succession. This will lead to frequent buy and sell transactions and is referred to as whipsaw.  The idea of taking a moving average over 10 months (an not over a shorter duration) is to reduce this effect.

There is nothing magical or special about 200 days or ten months. You can choose 12 months of 15 months and expect similar results.

Origin of the 10 MMA

Tactical asset allocation (TAA) based on the 10 MMA was popularized by PMS founder and CIO Mebane (Meb) Faber via his paper A Quantitative Approach to Tactical Asset Allocation. This is an easy-to-understand read and I strongly suggest that you try.

Faber constantly updates this strategy via his website. You can also get more details and analysis via extradash. I have also critiqued this strategy before: Is it possible to time the market?

With respect to the US market (which is what I have looked at, but it has been applied to many asset classes and international securities), the method almost always lowers risk with or without higher return. In what follows, I apply a simpler version of the method.

Rules of Market Timing with Ten Month Moving Average

We will compare two portfolios over a ten year investment period.

1: systematic investing into a portfolio of 50% equity and 50% fixed income. An amount of Rs. 1000 will be invested into the Sensex and an imaginary fixed income fund that offers a steady 6% return (this is considered to be a post-tax return). The portfolio will be rebalanced once a year. To account for exit loads and tax associated with this, the final portfolio amount will be reduced by 4%.

2: Tactical investing with the following rules.

Check Index once a month (at the start)

If current price > 10MMA (last ten month price average): Invest Rs. 2000 that month in equity(Sensex). Also, sell fixed income and invest into equity.

The fixed income sale and equity investment are considered to be done on the same day (which is impossible unless they are held as ETFs).

If current price < 10MMA: Invest Rs. 2000 that month in fixed income. Also, sell equity and invest into fixed income.

At this point, you may wish to compare these rules with those used in Market Timing with Index PE Ratio: Tactical Asset Allocation Backtest Part 1. You will notice an important difference.

In the PE-timing model, there was a window when the index PE was neither “low” nor “high”. During these periods, equal amounts (Rs. 1000) was invested in equity and fixed income. Such a window does not exist with the 10MMA. Either you invest Rs. 2000 in equity + amount redeemed from debt or you invest Rs. 2000 in debt + amount redeemed from equity.

The final tactical portfolio is reduced by 20% (this is 5 times the amount assumed for the systematic portfolio as the average no of trades in 10Y is about 5). This 20% accounts for exit loads of equity and fixed income and tax associated with equity. Remember equity was taxed like a debt fund before 2004 and will be again taxed from the current FY. The tax associated with fixed income is already accounted for in its returns (constant 6%).

At first sight, 20% may look a bit much, but you may change your mind when you see results below. Anyway, it is better to err on the side of caution. I hope these assumptions are reasonable. A market-linked fixed income security would have been better, but let us work with these for a start. Pre-tax returns of 8%+ were easily attainable up until 2014 or so. Therefore I think 6% post-tax is reasonable for the backtest.

Market Timing with 10 MMA: Results

In what follows, I also include the PE-timing results for comparison.

Portfolio Returns (XIRR)

The systematic and 10MMA portfolios were tested for 338 past ten-year periods and the PE model was compared for 209 periods with same investment dates. Both models were successful in getting a higher return for some periods. I do not want to get into numbers here (how many runs were successful) as that is often misinterpreted as a probability.

No of months underwater

Next, we consider the total no of months the portfolios were underwater. That is the no of months the portfolio was lower than a previous peak. This is known as a drawdown. Here is an example.

Notice the drawdown points on the left. The first time the portfolio dips from a peak the drawdown is -1% (non-zero). So that is the first month underwater.  We calculate the total number of such months the portfolio was lower than a previous peak (underwater)

Clearly, the index PE method is better (for the period studied). It has lower falls from a peak (I am referring to the portfolio here). The systematic method is typically better than the 10MMA too.

No of months continuously underwater

Now consider this example. Notice that within the rectangular area, the portfolio was always lower than its most recent peak. So the drawdowns (falls) calculated each month (red dots) within the box were negative. This means the portfolio was continuously underwater for 11 months.

So we ask, during each run, what the most no of months the portfolio was continuously underwater.

The PE method is better here too.

Maximum fall from peak (max drawdown)

So we next ask during each run what was the biggest fall from a pealk (drawdown)

Generally, both timing models had steeper falls (the vertical axis is negative). A -40% drawdown means a -40% fall from a peak from one month to the next.

Volatility comparison

This is the average spread (standard deviation) in the month to month portfolio returns. Higher the value, higher the portfolio ups and downs.

Since the equity allocation swings from 100% to 0% in both timing models, we can expect higher volatility. There is not much to differentiate the two models.

No of equity trades (sell-offs)

The 10MMA strategy seems to requires more equity sell-offs than the Index PE model.  Though both are reasonable. In spite of this, I have used the same 20% discounting factor for both models. If we recognise that 10MMA could entail higher taxes and loads, the PE model (at least the one considered) would be a better bet.

Months of investment in equity

Since the 10MMA model is a bit extreme when it comes to equity invest (either all or nothing) as mentioned above, it results in a higher no of months with no equity investment. Behaviourally it requires more discipline.

Max continuous period of no equity investment

Amusingly here the 10MMA is a bit more steady with the max period of no investment between 10 and 20 months. The PE model requires long stretches of no investment.

Summary

Did you come up to this point sequentially? Then thank you. Let me know if any of the above is not clear and leave your comments below. If there are any other strategy that you want tested, I will be happy to try.

Did you come straight to this without bothering to study any of the above graphs? Then, sorry, I will not cater to lazy readers and give a nice, “this is better than that” summary. Go back up and figure that out yourself or ignore this article altogether.

Remember Market timing needs a whole lot more discipline than setting up a SIP. Market timing is therefore pretty boring!

Will I adopt this strategy? No, because I don’t need to. I can manage risk with my goal targets. Also, I cannot stay away from equity investing for this long.

Should you adopt this strategy? Not my problem. You will have to figure it out.

Does this strategy work? The backtest says yes but keep in mind the assumptions made, real-life limitations, the fact that in the recent past results are not that good and that it is only a back-test.

Note: I will not answer any personalized question regarding this strategy. Also, the excel sheet used will not be shared. I consider it a proprietary research tool and will not share it until I milk it dry.