<--- GO BACK TO MARKET TIMING MAIN PAGE

Another very powerful momentum market breadth indicator used by the professionals is what we call the Bullish Percentage Indicator (BPI). This is merely a daily index that records the percentage of JSE shares trading above their 50-day moving average. This is intended to give us an idea how overbought or oversold the general market is.

This is a useful metric because as stocks move above and below their moving averages, they get stretched away from the average basis that traders or investors entered at. By looking at this you get an idea of how ‘healthy’ a market is. But it can also show you important inflection points in the market. For example, when almost all stocks in a market are below their 50 day moving average, you can bet that a snap-back rally is not too far away.

As nobody maintains such an index for the JSE we had to create one ourselves. We compiled this index going back 11.5 years (it obviously took quite some time) with the results shown below:



We see that the BPI, depicted by the brown vertical (and spiky) bar chart, oscillates between 5% and 80% (the maximum readings recorded for the last 11.5 years). We have determined through our own calibration efforts over 12 years that readings above 70% are overbought and readings below 30% are oversold.

AN OVERSOLD TIMING SIGNAL
You can see that when the BPI is less than 30% and reverses direction upwards, this is an EXCELLENT signal for market troughs (i.e short term market reversals). You can see this has happened 20 times in the last 11.5 years (every 6-7 months on average) as depicted by the green vertical bars. Only on five occasions (10/2000, 08/2002, 03/2003, 07/2008, 11/2008) did the BPI trough fail to identify an absolute ALSH trough, giving a success rate of 75%. The BPI oversold signal doesn't catch EVERY market trough, but you can be sure when it fires that you have a high confidence event. Whilst all BPI troughs coincide with local JSE troughs, only those below the oversold line have the accuracy upon which to make high confidence trades.

AN OVERBOUGHT TIMING SIGNAL
You can see that when the BPI goes above 70% and then makes a peak and reverses direction back down again that it is also a VERY GOOD signal for market peaks. This has happened 18 times in the last 11.5 years but the signal was correct only 12 times or 67% of the time. Most of the visible overbought signals are shown by the yellow vertical bars. Whilst every BPI overbought signal did indeed result in some contraction in the JSE, the 6 "wrong" signals merely got you out the market too early (as opposed to them being wrong and losing you money). This characteristic of a breadth indicator having better accuracy with picking bottoms than tops is common and is as a result of higher volatility experienced with sell-off's (which are panic driven) than rises leading to peaks (which are gradual and driven by greed and complacency).

USING BPI IN A TIMING STRATEGY
As you can see from the chart, not every crash is preceded with an overbought warning (just look at 2001 and 2007), but every overbought signal leads to a correction (some small, some severe). This means we cannot rely on the overbought signal alone to get us out the markets in time. However, the BPI, like most breadth indicators DOES provide an excellent warning for impending crashes - through DIVERGENCE.  When the JSE is making new highs, but the BPI fails to reach overbought territory (look at 2007 on the chart again) or is on a downward trend, you can be sure big trouble is looming. This is because stock market gains are being attributed to less and less shares (weakening market).

The oversold signal is more robust and quite suitable for picking JSE entry points provided you can time the BPI reversals below 30 or even 20. Using simple rules such as looking for a 10%-15% increase in the BPI from previous day on the back of price strength on the ALSH, and ensuring both ALSH and BPI display higher highs and higher lows enables you to be about 85% accurate in achieving this. Also the lower the BPI the more your chances increase of getting close to the trough. An example is shown below for the trough of the great 2008 crash where using the BPI would have allowed you to pick the trough of the bear market within 6-day accuracy (we looked for a greater than 10% one-day rise in the BPI in below example)



However, a far more accurate picker of JSE troughs is our TroughFinder Indicator. Whilst 95% of the BPI oversold areas correspond with TroughFinder signals, TroughFinder is far more accurate in detecting the exact turnaround points. We use the BPI oversold areas to confirm TroughFinder signals, but not to initiate them. To this end the BPI is probably more useful for picking out JSE peaks (through divergence)  and will be incorporated into our upcoming PeakFinder Indicator.

The PowerStocks BPI Timed Trading Strategy
Despite the minor issues discussed above we devised a rather successful mechanical trading strategy using the BPI and a fast 20-day moving average for the BUY signal line (as bounces off the bottom are more rapid) and a slower 50-day moving average for the SELL signal line (as market peaks take longer to appear). It goes as follows:

1. Only consider a BUY signal when the BPI is <= 30 (oversold conditions)
2. When the BPI crosses up from below through its 20-day moving average, then BUY
3. Only consider a SELL when the 20-day BPI moving average > 50-day BPI moving average
3. When the BPI crosses down from above through the 50-day moving average then SELL 

It's as simple as that! Assuming a brokerage fee of 0.75% per transaction, prime less 4% for the daily interest rate, then the above strategy would have traded 19 times in the 11 years since 30 June 1998 and only have been vested in the JSE for 31.4% of the time. It would have returned 714% growth versus an ALSH buy-and-holds' 277% growth. For the small amount of time this strategy was in the JSE this is a remarkable performance. For the 42 months this strategy was invested in the JSE (ie excluding the periods it earned interest in the bank), it returned 425% growth
 or a stunning 60.7% compound annual return.

A total of 14 of the 19 periods (73.7% of trades) produced a profit, with the minimum profit being 5.7%, just enough to cover the 1.5% brokerage fees. The 4 losing trades were contained to an average of -4.4% a characteristic designed into the strategy DNA with the 50-day moving average SELL line that will always limit losses. Average trade duration was 90 days with average CAGR per winning trade of 71.7% Those are very good odds and great returns. In addition, only 6 trades (31%) showed a loss within 5 days which makes this BUY signal reasonably accurate for short-term options/warrants/futures trades that are geared, as the strategy displayed a 69% chance of generating an immediate profit. The trade history is shown below:



There were 214.2 percentage points gained in the winning trades and 22.1 points lost in the losing trades to give a win/loss ratio of 9.69. This means that for every percentage point we lost, we gained 9.69 percentage points!

The chart below shows the total return index (TRI) of the BPI Timing Strategy versus an ALSH buy and hold strategy. The straight, slightly upward moving sections of the blue line are the periods the BPI timing strategy held money in the bank, earning prime less 4%. Note that like the JSE SwissClock (and most trading timing strategies as opposed to investing timing strategies) this strategy traded aggressively through the 2008 crash and got suckered into 2 bear market rallies where it lost 7.3% and 6.7% respectively. However the short 50-day moving average SELL signal line contained the losses. It picked up the March trough beautifully though, but this time the sensitive sell signal line got it out on 15 June for a 14% profit. As of this date it remains out the market.



The vested periods of the strategy, in relation to the actual BPI oscillator and the ALSH index is shown below. Click on the image to enlarge and see the 20-day BUY (green) and 50-day SELL (black) signal lines as well.



A weekly updated JSE Bullish Percentage chart appears in The Weekly JSE Pulse, together with the BUY and SELL signal lines, for our paid subscribers, as shown below. As usual subscribers get alerts when thresholds are exceeded in the HeadsUP! blog.
 


TIMING STRATEGY VARIATIONS
In our research we discovered a number of variations of this strategy that may suit different types of active traders. The main parameters that have an effect on the strategy are the oversold threshold used for triggering BUY signals and the length of the BPI moving average for the SELL signal. The BUY signal 20-day moving average is at its optimum - increasing it gets you in too late and decreasing it too early.

LINGER LONGER - We found that extending the sell signal moving average from 50 to 80 days also performs rather well. This strategy traded  17 times and delivered 728.23% total return whilst being vested in the JSE for 36.9% of the time. The trades were the same as that of the core strategy except that two of them were consolidated into other trades . So same trades but longer durations. This variation is a little more accommodating for minor corrections at the top before pulling the SELL trigger.

ALL YOU CAN EAT - This  cranks the BUY trigger threshold all the way up to 70% and works with the 80-day BPI moving average for the SELL line. Buying into the JSE when the BPI is above 70% is not recommended as you start buying into "heavily oversold conditions" when correction (or reversion back to the average) is more likely. But this aggressive strategy remains in play near market tops even when things start getting frothy. This strategy returned 609.65% with 28 trades (one every 4.8 months) and was vested in the JSE 59.0% of the time.


<--- GO BACK TO MARKET TIMING MAIN PAGE
 
Make a Free Website with Yola.