Seeking Alpha
Research analyst, newsletter provider, portfolio strategy
Profile| Send Message|
( followers)  

The idea of market timing is to get out of an asset class when an indicator or a combination of indicators enters a risky zone. Market timing is generally built on technical indicators. Sophisticated aggregate indicators may also be used. Some examples:

  • Put/Call ratio for options on an index derivative or a set of stocks.

  • AAII Sentiment Survey.

  • Number of S&P 100 stocks above their 200-day moving average.

  • Ratio of bullish/bearish chart patterns in a set of stocks.

  • EPS previsions for S&P 500 companies.

But the most popular market timing indicators use simple moving averages (S.M.A.) on 20, 50 and 200 days. The idea is to consider that the market is bullish when the price is above a particular moving average, and bearish when it is below it. A variant is to consider that the market is bullish when a short S.M.A. is above a long S.M.A. and bearish when it is the contrary.

Here is an example, with SMA50 and SMA200 on the S&P 500 index chart (from Yahoo Finance):

(click to enlarge)

We are looking for strategies on ETFs, so let's see what happens if:

  • We are invested on SPY (SPDR S&P 500 ETF Trust) when the price is above the red line (SMA 200) and out of the market the rest of the time.

  • We are invested on SPY when the green line (SMA50) is above the red line (SMA200) and out of the market the rest of the time.

This table compares the total return, compound annual growth rate (OTCPK:CAGR) and maximum drawdown (D.D.M.) of 13-year simulations (1/1/2000 to 1/1/2013). The condition is examined daily and trading fees are not taken into account.

SPYTotal ReturnCAGRDDM
Buy & Hold25.921.79-55.42
price>SMA20040.832.67-24.23
SMA50>SMA200109.255.84-16.96
SMA50>SMA200 leveraged x2274.9810.7-31.7

The detailed simulation report shows that the market timing based on two SMAs not only gives a better result, but triggers a lower number of trades. It also shows that the major advantage of being out of the market when SMA50 is below SMA200 is less to increase the return than to avoid the periods with the highest volatility and drawdowns. It may open the possibility to leverage the position.

The laziest investors may be seduced by a strategy based on the last line of the table. About 9% a year above the "buy-and-hold" benchmark on average, with a much lower historical drawdown, and only 15 trades in 13 years: it looks not so bad.

If you want to invest on this idea, keep in mind the following points, ranked by increasing importance:

  • The CAGR since January 2009 (13.69%) is below a non-leveraged buy-and-hold (14.96%). Market timing rules often gives false signals in trending markets.

  • The Sortino ratio is low: 0.3 (this is my preferred Sharpe-like ratio because it ignores the "good" volatility).

  • The Kelly criterion is 0.06: it tells us that from a gambler's point of view, one should not invest more than 6% of a portfolio on this strategy.

  • The longest historical drawdown on 13 years was about 40 months.

Such a rule is designed to avoid long bear cycles. In trending markets it may generate false signals and underperform the benchmark. It did since 2009. The good news is that even on this underperforming period, the drawdown and volatility are kept lower than a "buy-and-hold" position.

You may wonder what happens if you don't want to check the market timing condition every day.

The following table gives the returns and max drawdowns of the leveraged version, depending on the market timing frequency: every day, every week, every 4 weeks. For the later I have simulated four starting dates: 1/1/2000, 1/8/2000, 1/15/2000 and 1/22/2000.

SMA50/SMA200>1 leveraged x2Total ReturnCAGRDDM
daily274.9810.7-31.7
weekly288.0710.99-33.03
4weeks1228.439.58-37.61
4weeks2250.1610.14-39.51
4weeks3171.528.01-34.98
4weeks4176.298.17-36.87

The strategy keeps returns and drawdowns better than buy-and-hold on all time scales. The difference between a daily and a weekly checking is not significant, however downgrading to a monthly checking makes the results much more sensitive to the starting date and degrades significantly the drawdown and/or the return.

To make the numbers more realistic, I propose to apply a 2% borrowing rate for the leveraged part, and a 0.1% rate for trading fees and spread. Then in the case of a weekly checking, the CAGR is 9.4% and the drawdown 34%.

You may object that 13 years is quite a short period in the financial market history. Are we sure that this market timing rule really works? Various studies suggest that the results of a SMA200-based market timing are better on the very long term than between January 2000 and January 2013.

Here are numbers (S&P 500 index including dividends) from 1900 to 2008 borrowed from "A Quantitative Approach to Tactical Asset Allocation" (M. Faber 2009):

S&P 500 w/div.CAGRDDMVolatilitySharpe
Buy & Hold9.21-83.6617.870.29
Market Timing10.45-50.3112.010.54

The hypotheses are slightly different: Faber's timing is monthly and his condition is that the price must be above the 10-month moving average (which is almost equal to SMA200 in a daily time scale). In fact there is a high probability for the results to be better with a double moving average and a weekly checking.

These numbers suggest that on the very long term the CAGR is above 10% a year without leveraging. They also suggest that leveraging is not bulletproof: it would not have survived the 1929 crisis and a drawdown of 50%.

So if leveraging is not reasonable with pessimistic hypotheses, why bothering with market timing to make only 1.24% a year above "buy-and-hold"? The answer is in the drawdown. If we know a 1929-like crisis, you might be in a professional or personal situation requiring to withdraw some money from your position. It would be much easier if it is down "only" 50% from its highest point, not 85%.

And in the case you are optimistic and lose your position by leveraging? If you have listened to John Kelly's advice, it is limited to less than 6% of your investments.

The next part will give statistics about more indexes, sectors and other asset classes. Follow me if you don't want to miss it and Click here for more information about our research.

Source: Does Market Timing Matter? The Case Of SPY (Part 1)