OTHER TIMING MODELS--> | Repo/Prime Rate Model | Trendex Model | TrendexPrime Model
                                        |SUPERModel Composite Timing System |

We now examine a more sophisticated market entry/exit system based upon the South African Reserve Bank Composite Leading Economic Indicator Model.

The SARB has been maintaining and publishing this model since 1983. It measures 13 leading economic indicators in an attempt to predict the economic cycles 6-8 months in advance. Backtests by SARB have shown that this model fairly accurately marks top and bottom turning points in the economic business cycle. Since stock markets are "forward looking" by 6-8 months we concluded this might be a good model for timing JSE entries and exits.

The Composite Leading Economic Indicator is published monthly on the SARB web site and runs 2 months in arrears. So in June 2009 they will publish a reading that represents April 2009. The indicator is compiled from the following data:

  1. Opinion survey of volume of orders in manufacturing
  2. Opinion survey of stocks in relation to demand: Manufacturing and trade
  3. Opinion survey of business confidence: Manufacturing, construction and trade
  4. Composite leading business cycle indicator of major trading-partner countries: Percentage change over twelve months
  5. Commodity prices in US dollars for a basket of South Africa’s export commodities: Six-month smoothed growth rate
  6. Real M1 money supply (deflated with the CPI): Six-month smoothed growth rate
  7. Prices of all classes of shares: Six-month smoothed growth rate
  8. Number of residential building plans passed for flats, town-houses and houses larger than 80m2
  9. Interest rate spread: 10-year bonds less 91-day Treasury bills
  10. Gross operating surplus as a percentage of gross domestic product
  11. Labour productivity in manufacturing: Six-month smoothed growth rate
  12. Job advertisements in the Sunday Times newspaper: Six-month smoothed growth rate
  13. Opinion survey of the average hours worked per factory worker in the manufacturing sector
The historical readings of this indicator are shown together with the ALSH index below. A visual inspection immediately confirms that this indicator (the BLUE line) seems to have remarkable accuracy in coinciding with JSE bull and bear market start/stop points. Note the ALSH index (RED line) is logarithmically scaled to accommodate the long time period and allow you to see all the peaks and troughs of the JSE.

We note that this model accurately timed the commencement of all but one large crash. It showed divergence to the stock market in  1988 to 1992 and 1995 to 1998, both periods where it told us to "stay out" the market. The first divergence meant we missed the massive JSE recovery from crash of '87 and the 2nd divergence kept us out JSE market growth but ultimately saved us from the crash of '98. This strategy is the only one we have seen so far that kept us in the JSE all through 2007 just before the great 2008 crash.

Although the above chart looks very promising, it has the advantage of 2 months worth of "hindsight" due to the data being published 2 months in arrears. So we need to modify the above chart by "time slipping" it by 2 months to give us a view on what data an investor would realistically have had at their disposal in order to make market entry/exit decisions.

Even with the 2-month slippage, the curve fit seems very good. The only major downfall here was it got us into the '87 crash and kept us out the first 60% of the subsequent recovery. This is likely to severely impair any total returns on a backtest of this strategy, so keep this in mind.

Even with time-slipped data, it is not realistic to be able to use the above chart to time your market moves - because how will you know if you are at the peak or trough of the indicator at any point in time? You could wait another 1-2 months to observe a turn, but now your time slippage extends from 2-3 to 4-5 months meaning we don,t have much of a "leading" indicator anymore! In addition, if you observe the BLUE line in the above chart you see there would have been many "false turns" (both up and down) which would have killed the returns on any investment strategy.

What we need is an OSCILLATOR linked to the Indicators' rate of change to act as a BUY/SELL signal line. We tested many periods for rate of change and found that the 12-month rate of change ( since the same reading last year) offers an excellent oscillator (with some modifications) for "mechanising" this strategy.

The timing strategy for the PowerStocks Econometric Timing model is centred around the customised oscillator above and works with 6 very specific fixed mechanical rules and a 2nd derivative calculation to time the oscillator troughs to generate highly effective BUY and SELL signals.

Using the above timing would have delivered 45,570% growth over 31 years, or 22% compound growth per annum. This is in contrast to the 8,016% growth (15.35% compound) delivered by a JSE ALSH buy-and-hold strategy. The transactions of this strategy are depicted below:

The PowerStocks Econometric Strategy would have triggered 7 buys and 8 sells in 31 years. Average time in the JSE is 30.86 months and average time out the JSE was 21.86 months. You would have been in the market for 58.5% of the time, and 41.5% of the time in a bank or money market account. Average growth for the vested periods in the JSE was 94.5%.

Average growth of the JSE whilst the strategy had us "avoiding" the market was a paltry 4% with a frightening 19.5% standard deviation. During these "avoid the market" periods we would have averaged 31.9% interest fuelled growth! This really had a knack for keeping us out the JSE during lacklustre periods.

A comparison of the Econometric Model (referred to below as the "Reserve Bank Model") to other strategies we have covered previously are shown below:

The total returns of this strategy versus a ALSH "buy-and-hold" strategy are plotted below:


Not bad at all - this strategy delivers similar returns to the Repo Rate (Prime Rate) Model we covered previously, but note its time in the market was longer and there were 8 transactions with this strategy versus 12 with the Repo-Rate model. It was also in the market for different periods than that of the Repo-rate strategy. For example, the Repo-Rate strategy bailed out the markets with the first interest rate hike on 08 June 2006, whereas the Econometric model rode that massive bull run all the way to September 2007 (8 months before the great 2008 crash.)

Over the last 10 years, the Econometric model (SARB Model in the below chart) has been our top performing strategy, mainly due to its tenacity in holding onto that last bull run up until the last minute, when all the other strategies had bailed out into cash. However, over 20 and 30 years, the strategy lags somewhat (see the SuperModel report) so you are discouraged from using this model alone in investment decisions as it is unlikely we will see a bull run like 2003-2008 again. Our composite SUPERModel timing system has been designed to cater for many more market/economic situations than this model and  delivered 4 times the performance over 30 years.

From our previous research, we now know that expansive monetary policy (falling interest rates) have a powerful effect on share prices and you have just seen from the Econometric model how the state of the economy also has a powerful effect on share returns. But the two are mutually exclusive. Sometimes there are periods when the economy is growing but restrictive monetary policy conditions are at play (such as when the economy is overheating), and conversely, many periods exist when the economy is struggling (recession) but expansive monetary conditions are at play (ie interest rates are dropping to try and stimulate the economy.)

So naturally it makes sense that a timing model that combines the Repo and Econometric timing signals should deliver superior returns. When both models are firing a BUY signal we have a favourable interest rate environment coupled with a growing economy that should turbo charge the JSE. Similarly, combining the two timing models should mean we receive more BUY/SELL signals than when just using one model, which should give us more time in the JSE and hopefully larger returns.

We tested a strategy that was invested in the JSE when EITHER of these methods were presenting a BUY signal, but this only came in at 33,000% growth by being vested in the JSE for 77% of the time. No slouch when compared to an ALSH buy-and-hold strategy but not so interesting for us since it fails to outperform either one model. Besides it is too long in the market (we're looking for highest growth for shortest time in the JSE) meaning that it has higher risk than the other strategies.

So we back-tested a more stringent strategy of only being invested in the JSE when BOTH these two strategies were giving a BUY signal, under the assumption that this should yield periods of "turbo charged" JSE returns.

The results are very interesting - we boosted returns from 45,510% to 59,876% whilst reducing our time in the market to a stunning 36.4%. So we have far higher returns with less JSE exposure (and hence less stock market risk). This boost in returns is similar (but not as extensive) as that we achieved when combining the Repo-rate model with the Trendex Model (to produce the TrendexPrime model)


Before you get all excited and think we have hit pay-dirt - have a glance at the bottom of that table to the item marked "SUPERMODEL". Did your eyes pop out their heads? Well that is what we have been working on for the last 6 months and leading up to with all our recent research.

The SUPERMODEL Strategy is a combination of the Econometric (Reserve Bank)Model, the Repo Rate (Monetary Policy) model and the Trendex (Momentum) model that delivers on our original mission statement 12 months back : "To produce a JSE market timing strategy that exceeds 25% CAGR for the last 31 years, consistantly avoids bear markets and JSE crashes, and does so by being vested in the JSE for no more than 50% of the time."

As a PowerStocks subscriber, you get access to weekly updated timing signals generated by the SuperModel. You can use these to structure your own "Timed ETF" investment strategy by using the signals to buy/sell Satrix-40 or some other ETF or you can use them to structure purchases around your own BLUECHIP mini-portfolio (which delivered 150,000% growth with the SUPERMODEL) or just use the signals as a MARKET STRENGTH INDICATOR for your general share and trading purposes.

We publish a monthly update of the Reserve Bank Leading economic indicator as well as the Rate of Change Oscillator on the Weekly JSE Pulse page (subscribers only).
Make a Free Website with Yola.