Market Timing

Market timing is defined as “the strategy of making buy or sell decisions of financial assets (often stocks) by attempting to predict future market price movements.” The prediction may be based on an outlook of market or economic conditions resulting from technical or fundamental analysis.

Buy low, sell high. Easier said than done! Numerous flash studies have addressed the issue of market timing by quoting investment returns with and without some kind of market timing. Proponents of market timing often quote the wonderful returns one could have achieved by missing the 10/30/60 worst days, while those skeptical of market timing quote the opposite; the reduction in returns due to missing the best 10/30/60 days of any given bull run. In the real world however, no matter what type of investments one has, it is not possible nor practical to somehow miss a few days during a bull/bear market as disinvesting and reinvesting takes time and costs money (transaction costs). Even if one had instant liquidity, based on William Sharpe’s 1975 article entitled “Likely gains from market timing”, one needs to guess right 74% of the time to extract any benefit from market timing.

However, studies of market timing in the US¹ have focused on the effective use of signaling tools to make investment decisions: Shiller (2006) proposed using the level of the PE ratio, while in 2002 Pu Shen calculated excess returns of timing strategies based on the spread between the S&P 500 earnings yield and the 3m T-bill. Perhaps one of the simplest strategies is using moving averages (MA) to time market entry- and exit points. When the 50 day MA of the market moves above its 200 day MA, the trader switches to equity. When the 50 day MA goes below the 200 day MA, the trader sells equity and buys a money market fund. From January 1994 to January 2009, this strategy has returned 11.4% annually with a maximum drawdown of 19%. Only 14 trades were made in the 15-year period, an average of less than one trade per year. This compares very favorably with a buy-and-hold return of 6.6% and a maximum drawdown of 50.8% for the US equity market (ending in November 2008) over the same period. As a result of the signals, the 50/200 MA trading strategy switched to money market at the end of 2007, avoiding the massive market loss in 2008.

Replicating this study using the All Share Index from July 1995 to May 2009 yields surprisingly different results, with the buy-and-hold strategy outperforming the 50/200 MA trading strategy with a 14.5% annual return vs. 13.4% per annum. The portfolio was switched 17 times in the 14 year period and of the 8 times the MA indicated a switch to cash, equities outperformed cash 6 times. This highlights the negative impact that the MA lag (as much as 5 days using a 10-day MA) has on performance. Using an exponential moving average² (EMA) to reduce the lag effect produces slightly better results. The EMA trading strategy achieves an annual return of 16.1%, thus outperforming the buy-and-hold strategy by 1.6% per annum. This is more than enough to offset the small impact of switching fees caused by the 15 switch signals. It is important to note that the outperformance was primarily due to the strategy signaling a switch to cash in July 2008 thus avoiding further large falls in the market towards the end of 2008 and the beginning of 2009. As end point bias³ has a large effect on the results of this type of strategy, the ideal environment for a MA timing strategy is one where a large correction is followed by a slow trending recovery. This gives the strategy time to switch out of the cash position (which was triggered by the correction) and continue to benefit from the subsequent upward trend. This has not yet happened and the strategy remains fully invested in cash.

A graph of rolling 2 year returns (annualized) shows how the EMA market timing strategy protected capital (with a maximum drawdown of -27.6% vs. -45.4% of the ALSI) but lagged the market at times.

Chart 1: Rolling 2 yr annualised returns of EMA market timing strategy vs. All Share Index

 (Click on image to enlarge)

Source: Glacier Research, I-Net Bridge

Most of the US studies based on valuation metrics and signaling strategies have shown that market timing, if done correctly and without emotional interference, can add value to a portfolio over time. In South Africa however, investors need to identify more innovative investment strategies as our less efficient, highly volatile, commodity-driven market continually surprises with its resilience compared to its counterpart across the Atlantic.

While a lack of trading systems makes it difficult for the average investor to employ successful market timing strategies, using predetermined “trigger levels” to adjust existing portfolios can be practiced by investors. Here, one adjusts the portfolios asset allocation when the equity component has risen beyond an acceptable level by trimming existing equity positions (taking profits) while adding to cash reserves, or to an asset class potentially offering more value.  

Rebalancing

And what about rebalancing? Rebalancing is the periodic adjustment of a portfolio to predetermined levels, based most often on a strategic asset allocation. The main difference between rebalancing and market timing is that where market timing attempts to predict future market movements, rebalancing is pre-determined.

Examining the effect of periodic rebalancing on a typical balanced portfolio (50% SA equity, 15% foreign equity, 10% bonds, 10% property and 15% cash) over the past ten and twenty years, one can conclude that the difference in end values is negligible for all but the portfolio that was never rebalanced.

Table 1: Annualised returns of a typical balanced portfolio with different rebalancing frequency

Source: Glacier Research

One may be tempted to forgo portfolio rebalancing to maximize returns, but it is likely that the risk of this portfolio (that is never rebalanced) will exceed the investor/fund’s risk limits over time as the weighting to asset classes with good relative returns increases. The table below shows the variation in risky exposure (equity + property) from the strategic (initial) allocation (SAA) of 50% equity and 10% property that occurs over time with different rebalancing frequencies.

Table 2: The ranges of risky exposure (equity + property) with different rebalancing frequency

Source: Glacier Research

While an average exposure to risky assets of 65% remains close to the SAA, a range of 59% to 83% in risky assets is perhaps too large for a typical balanced mandate and at some stage equity and/or property exposure needs to be trimmed to align the portfolio with its strategic allocation.

A tolerance-based approach to rebalancing has the potential to generate excess returns without the need for a contrarian view and can be applied by investors themselves. This simple strategy involves selecting an appropriate strategic asset allocation based on risk tolerance, financial and liquidity needs as per the financial planning process. This SAA is then only rebalanced when any of the risky asset class exposures exceeds a certain predetermined level, say 5% more/less than the original allocation. This ensures that the portfolio remains consistent with the investor’s risk tolerance, while at the same time ensuring that assets that have appreciated markedly are reduced back to the original allocation. One could surmise that this effectively as “buy lower and sell higher”, which is after all one of the best ways to make money.

A US study by Phil deMuth using 10000 Monte Carlo simulations on a seven-asset-class portfolio states that “the more frequently we rebalance, the worse our returns”, while “tolerance-based approaches, on the other hand, seem to have more merit; those portfolios that were allowed some leeway to roam before being rebalanced had better risk-adjusted returns than those adjusted according to the calendar.”

Simulating returns in a South African context⁴ leads to similar conclusions, with the tolerance-based approach providing better risk-adjusted returns than the periodically rebalanced portfolios. Periodic rebalancing leads to sub-optimal adjustments to the portfolio as it does not allow for sufficient growth to occur before weightings are re-aligned with the original allocation. Allowing for some growth (within risk limits) before taking profits allows one to take advantage of momentum effects that are often ignored by asset allocation strategies and ultimately leads to superior longer term performance.

Chart 2: Simulated risk-return trade-off of different rebalancing strategies

 (Click on image to enlarge)

Source: Glacier Research

Over the past 20 years, applying this strategy (5% tolerance to limit risk) to an actual South African balanced portfolio (50% SA equity, 15% foreign equity, 10% bonds, 10% property and 15% cash) would have generated an excess return of 0.8% per annum over a buy-and-hold strategy; and 0.6% per annum over a static allocation that is set to the SAA monthly. These numbers may not seem significant, but over 20 years, 0.8% per annum on an original investment of R1,000,000 equates to an extra R3,387,507 at slightly lower levels of risk than the buy-and-hold strategy. And with only 20 rebalancing triggers over the 20 year period, transaction costs are minimized.

Table 3: The effect of rebalancing strategies on a South African balanced portfolio

Source: Glacier Research

The example using actual data mimics the results of the study using simulated returns with the tolerance-based approach yielding better risk-adjusted returns.

While pure market timing is difficult to implement successfully, slight adjustments to a portfolio (whether predetermined or tolerance based) can have a considerable impact on investment values in the future and should be included as part of an investors financial plan.