GO TRENDEX --> | 31 Year Performance Backtest | See this weeks JSE Trendex reading |
                         | The Combined Trendex/Repo Rate Timing Model |

Below is a chart of the JSE ALSH Index over the last 33 years, together with the Coppock Indicator Curve (otherwise known as the Trendex Curve) and the 9 BUY signals that have been flashed.

As in overseas markets, the Trendex Indicator has provided remarkable accuracy on signalling the beginning of Bull markets on the JSE. The table below shows the 8 times (excluding the one that just fired on 20 May 2009) the Indicator has triggered a BUY signal over the last 33 years, together with the subsequent 1,2,3,...12 month growth in the ALSH index. 

We see how rare and significant these BUY triggers are. The one instance where the trigger gave a wrong-way trade was on 1998/02/26, just before a fat crash and you can see from the table above that a negative result ensued right through all the periods from 120 to 360 days. However we wish to note that if you look at the very top chart, the BUY signal (Number 6) was VERY WEAK as it was not very far below the zero line.

Looking at the above table, you see that if you buy on a trigger, you can expect an average 6.9% return after only 30 days, with a tiny 4.2% deviation. Even better, you can average 10.1% return after 60 days, with a small deviation of 5.6%. But the best risk adjusted return (Sharp of 2.26) is delivered after 90 days with 9.9% average return with only  4.4% standard deviation. The "ANNUAL" row shows us that this equates to a 45.6 annual rate of return! Historical accuracy of the signal for 90 days is 100%.

Could we use Trendex to time entry as well as exits in the market? As we have noted, Trendex is excellent at getting us into large bull markets after a heavy decline, but its signals at the top are prone to "false alarms". Look at the Trendex curve right at the top of this page and you will note that above the signal line, (during market advances) the curve flattened out and even declined sometimes, before resuming its upward march, leading to false SELL signals. This never happens with declining markets, hence the accuracy of the BUY signal.

For our Trendex Timing Strategy, we will  IGNORE all BUY signals when the Trendex score was not less than -5%, to ensure we are in a "significant enough" decline and have a "strong enough" signal. That would basically have ignored SELL signal #6 on 26/02/1998, the ONLY wrong way trade predicted by Trendex in the last 33 years.

To use Trendex for SELL signals we will have to try and time the top peaks as close as possible. We found it possible to time these peaks by looking for those occasions where the daily Trendex score measured greater than 50% and was 5% less than the maximum Trendex score for the last 30 days. So for example, if the Trendex score for the day was 50%, and the maximum score recorded for the 30 days prior was 56% then we would trigger a SELL signal. Whilst not timing the absolute peaks perfectly (which is impossible unless you can predict the future) we came exceptionally close on all occasions. More importantly we picked up EVERY peak and NO false signals!

Using the above BUY/SELL mechanical rules to enter/exit the JSE and during periods we are "out the market", sticking our funds in a bank account to earn the interest of the day (prime rate less 3%), we calculated the returns over time as shown below:

We see that the Trendex mechanical investment strategy actually fared better (55,226% growth) than the strategy using Interest Rate Cycles to time market  entry/exits (45,570% growth over 32 years). Some interesting stats on the various strategies are shown below:

Do you see what we see? Look again, before reading further! Did you see it? The Trendex/Coppock strategy was only in the JSE 39.5% of the time! The periods the strategy was in the JSE returned 78% growth on average versus the 16.7% the JSE was delivering while our money was safely tucked away in a fixed interest investment.

In PITBULL Ver 2.0, we combine the Interest Rate Timing method with the Trendex Timing method to produce a combined strategy that delivered a stunning 65,000% growth over the last 32 years, by only being in the JSE 50% of the time.

On 19th May 2009 (signal 9 on the charts), the Trendex curve flattened its decline and made a reading of -59.6%. On 20th May it read -59.5% and on the 26th it read -59.4%. By 3rd June it read -59.0 confirming it had made a turn. The indicator very, very rarely makes false turns. The nature of the curve is such that when it turns up it STAYS up (the same cannot be said for the downward turns).

It is interesting to note that the indicator flattened out the day after the Repo rate was dropped by another 1%. In fact we have a perplexing coincidence, that we cannot explain yet, of the indicator turning up within less than a 7 day window of an interest rate drop over the last 25 years! And in another extraordinary coincidence, every initial interest rate drop after a period of rises (one of the main contributing triggers we use in PITBULL), occured within 5 days of a Trendex upturn! More on that later...
Make a Free Website with Yola.