RELATED TOPICS --> | Large resolution historical BITS Signal chart |
                             | The Martin Zweig BITS Signal variations |

The Breadth Impulse Trading Signal (BITS) is a powerful trading signal developed by PowerStocks Labs for use on the JSE that deploys daily advance/decline data to generate infrequent (every 6-10 months on average) BUY signals of very high confidence (85-87.5% win rate) and excellent win/loss ratios (in excess of 13 percentage points gained for every point lost) that can be used by medium term traders and long term investors alike. BITS is a tool the likes of which only sophisticated hedge fund managers normally have access to and gives our PRO subscribers a real leg-up in their trade risk management.

1. SIGNAL COMPUTATION AND USE
The BITS signal line is computed daily from the JSE's closing advance and decline data as follows:

BITS = EXPONENTIAL MOVING AVERAGE OF ADVANCES/(ADVANCES + DECLINES)*100

The length of the exponential moving average is confidential to preserve our intellectual property, but has been calibrated over the last 14 years to deliver the best results on the JSE. In addition, a special subset of JSE advance/decline data was used to only include main board shares, and exclude warrants, ETF's and interest rate sensitive issues such as debt. This is because the JSE is very sensitive to interest rates and the limitation of the shares to main board ones improved the performance of BITS considerably, and avoids "double counting" issues of  ETF's, SSF's etc.

A BUY signal is generated when BITS exceeds 60 and remains in force until BITS declines to 41 or 45 and below. There is no SELL signal, merely an expiration of the BUY signal. When the BUY signal is present we are in a BULLISH phase and when the BUY signal expires, we drop to NEUTRAL.

Long term investors will use a BUY signal as confirmation of a new bullish phase in the market to either keep holding onto their investments or "load up" their portfolios with more of their favourite shares and ETF's (allocate more capital to the share market.) They will ignore the "sell" generated by BITS. Based on the premise that a rising tide will float all boats, an expectation of favourable returns by the JSE index means that most (if not all) of the investors' favourite share picks are likely to benefit from the coming bullish period.

Medium term traders will use the BUY signal to climb into vanilla un-geared ETF's such as SATRIX40 or SATRIX-RESI or conservatively geared warrants and single stock futures (SSF's) that have the ALSH, TOP40 or RESI index as the underlying. Basically any investment instrument likely to mirror or proxy the ALSH index returns will be suitable. Contrary to the long term investors, they will EXIT these positions when the BITS signal declines below the 41-45 "SELL" zone.

A sample BITS signal chart as shown in the JBAR  JSE TIMERS sub-sheet in Chart Set 2a-2d is below:



2. THEORY AND BACKGROUND
BITS works on the premise that approximately every 6-10 months on average, there is a period of  breadth advances (termed a "breadth thrust" or "breadth impulse") that becomes so powerful as to jolt the market into a bull run. Achieving certain levels of positive breadth momentum suggests further market advance to come since the capability of breadth pulses to achieve defined, unusually high levels that are well-above average market breadth, more often than not signals the start rather than the end of meaningful market advances.

The phenomenon was first widely read about in Martin Zweigs' seminal "Winning on Wall Street" classic, where he measured 10-day rolling advance/decline ratios and tagged very rare instances when this reached 2-to-1 as powerful thrusts that "launch bull markets like a rocket ship." (we track a similar metric on JBAR chart 2a.) His reasoning was unless enough "thrust" in the form of advances leading declines is imparted to the markets, they may launch briefly higher, but soon succumb to gravity. He made an analogy of bull markets requiring enough initial thrust like a rocket, to provide enough momentum to escape the earth's gravity and reach orbit. Zweig went on to extend this theory with the formula now used by BITS, and he coined it the "Breadth Thrust" in which BUY signals were generated when the indicator moved from the "Sell Zone" to the "Buy Zone" within 10 trading days. These breadth thrusts were very infrequent however (one every 2-3 years on average) 

Another famous investor, Gerald Appel (inventor of the widely used MACD indicator) praised this indicator in his two best-sellers, "Power Tools for Active Investors" and "Beat the Market, Win with Proven Stock Selection and Market Timing Tools". However he modified its use in that he ignored the 10-trading day rule and removed the restriction that the indicator had to originate from the "Sell Zone" to generate a BUY signal. He called his indicator the "Breadth Continuation Impulse" and showed this to be very successful in timing winning trades on the US markets.

The thresholds and periods Appel used in calculating his indicator had to be extensively modified for the JSE, since our markets are so different to the US markets, but the performance characteristics of BITS closely matches that observed by Appel on the US markets over the last 33 years. In fact Appel's investment company, Signalert, use variations of this indicator in timing models for their investment clients. According to Appel, this is just about the best and most reliable quantitative short/medium term market timing method he has observed in his illustrious career. As you will see later on we will agree with him.

Appel states: "Signals generated by this indicator do not represent a total market timing model. They do isolate limited periods, which occur relatively infrequently, when the odds strongly favour at least short-term profitability in the stock market. Conclusions of entry signals do not, in and of themselves, indicate a stock market decline—only that the immediately positive influences of outstanding market breadth have been ameliorated."

3. SIGNAL PERFORMANCE FOR BITS VARIATIONS
Although the BIT signal is infrequent, it leads to very high confidence trades and magnificent win/loss ratios if you use the EXIT zone to terminate trades. The EXIT zone reflected in the above JBAR chart stretches from 41 to 45 depending on your trading style. The sections below refer to a 14 year back-test period of 5th March 1996 to 29 Dec 2009.

3.1 BITS-45
Using 45 to exit generated 20 signals since 1996 with a 85% win rate. Average wins were 7.15% versus average losses of only -3.16% to produce an average trade expectancy (gain) of 5.6%. A total of 121.57 percentage points were gained in the winning trades versus only 9.47 percentage points lost in the losing trades to yield a stunning win/loss ratio of 12.84 (12.84 points gained for every point lost over the 14 year period.) The full details of this trading strategy appear below:



You would have been vested in the JSE for only 29.1% of the time (4 years in total across all trades). Using the BITS system to trade the JSE and sticking your money in a money market fund or other instrument earning prime less 4% during un-vested periods would have returned 937% versus the ALSH buy-and-hold strategy which returned only 316%. BITS-45 thus achieved 2.6 times the performance of the JSE. The high SHARPE ratios show that the average win from each trade is relatively consistent.

3.2 BITS-41
Using 41 to exit relaxed the cancellation signal threshold and generated 16 signals with a 87.5% win rate, trade expectancy of 10.5% and an unbelievable win/loss ratio of 16.94! An investment strategy using BITS-41 to play the JSE and use money market accounts during un-vested periods would have returned 1,103% versus the JSE's 316% (The TRI in the tables refers to the return on R1.00). BITS-41 stayed in the JSE much longer than BITS-45 (43.1% of the time versus 29.1% of the time.) There is thus the argument that BITS-41 is exposed to more JSE risk than BITS-45 and therefore offers less risk-adjusted returns, but given the win-rate we do not think this is valid.



The higher SHARPE ratios show this strategy to offer less variability than BITS-45 in its average trade win and loss percentages which means it is more consistent in expected outcomes. The transaction log of the 16 individual BITS-41 trades with their entry and exit dates appear below. In all our calculations we took the conservative view that the investor would only act on a BUY/SELL signal the following day after it was published:



We note that vested periods average 133 days but have a wide range from as few as 14 days to as much as 362 days. If you use the exit signals to close trades, you will definitely have to treat these as trades and not investments from a tax point of view. Trades that gained more than 10% are shaded green and represent 56% of the total transactions. The average compound annual growth rate (CAGR) is shown as 31.3% per trade, but winning trades averaged 40.1% CAGR. Note there were three periods where you sat in cash for more than 438 days but due to the wide variation of waiting periods between trades it is not advisable to tie your cash up into very long call deposits as you may need it suddenly for the next trade.

3.3 BITS-ST (PowerStocks Short Term version)
Tom McClellan, son of the parents who invented the famous McClellan indicators, modified Appels' strategy with an exit rule that closed out any trade automatically after 50 trading days, provided the signal was below the BUY threshold. As per normal, if the SELL threshold was reached before this the trade would still come to an end. This provided for higher win rates but more and shorter trades. We found that a 41 day exit limit coupled with BITS-41 to work the best on the JSE as shown below:



BITS-ST will be a favourite with short term traders as it has many more trades (25 versus 16 for BITS-41 and 20 for BITS-45), a superb win ratio of 92% and still maintains a very respectable win/loss ratio of 12.23. What's more, trades are restricted to 41 days maximum meaning you get your profits out sooner! An investment strategy using BITS-ST to play the JSE and use money market accounts during un-vested periods would have returned a whopping 1,246% versus the JSE's 316% Also note how incredibly consistent the average win of 6.76% is - it only has a standard deviation of 3.27% to give an eye-popping SHARPE of 2.06.

3.4 BITS-MT (PowerStocks Medium Term version)
We liked Tom McClellan's' modification that allowed us to build a high performing version of BITS for our short term traders, and got to wondering if we could do the same for our medium-term traders. We found that setting a time limit of 90 days (3 months) on the trade to force a SELL signal produced very good results indeed:



An investment strategy using BITS-MT to play the JSE and use money market accounts during un-vested periods would have returned an astonishing 1,420% versus the JSE's 316%. The win-rate of 94% and the win/loss ratio and the TRI are the highest of all the variations we have looked at. The strategy managed this whilst being exposed to the JSE for 10% less time than the vanilla BITS-41. The trade transaction list for BITS-MT is displayed below:



BITS-MT is superior to BITS-41 in every possible way. It has one more trade than BITS-41 and outperforms it in win rate, win/loss ratio and 14-year total return. The variability in trade lengths is far lower with an average trade length of 97 days with a standard deviation of 40.5 days, to give a Sharpe of 2.4 whereas BITS-41 averaged 133 days with a wide standard deviation of 92 days to give a Sharpe of 1.44. Finally, BITS-MT had an average CAGR of 46.3% for its winning trades whilst BITS-41 had an average CAGR of 40.1% for its winning trades.

Note that although our vested periods are limited to 90 trading days, this works out to up to a maximum of 134 days in the above 14 year period since weekends and public holidays need to be taken into account and sometimes when the time-limit kicks in the BITS signal is still hovering above the BUY signal of 60 which will force the strategy to remain vested.

LONGER TERM TESTS
We only have detailed, reliable breadth data for the JSE going back to 1996 (14 years), but Gerald Appel did back-test the Breadth Impulse timing mechanism on the NYSE over a 34 year period from 1970-2004.

Twenty-six signals were generated between December 1970 and May 2004. Twenty-two (84.6%) of these signals proved profitable; four (15.4%) trades were unprofitable. The average gain per profitable signal was 5.24% per trade. The average loss per unprofitable signal was 2.37%. A total of 115.28% was gained during profitable trades, compared to a total loss of 9.5% during unprofitable trades. All in all, the gain:loss ratio came to 12.13:1.

These results compare very closely to those of BITS-45. Appel observes that many timing systems have waned in performance over time, except for this one which has remained consistently reliable.

SUMMARY
Although BITS-41 has an average trade length of 133 days, these range wildly from 14 days to 362 days! BITS-MT will limit trade lengths to 90 days and thus exhibit much less variability in trade length, whilst out-performing BITS-41 and providing for more trading opportunities with 10% less exposure to the JSE. BITS-ST is ideal for shorter term traders. No matter which system you deploy, one has to admit that they all perform exceptionally well. From a PowerStocks perspective our flagships from this timing strategy will be BITS-MT and BITS-ST which will be the systems we will trade with.

One could deploy multiple strategies in a funds phase-out mechanism by exiting 20% (for example) of your trade when 45 is reached (BITS-45) and a further 30% when BITS-ST issues a sell and a further 50% when BITS-MT issues a sell signal. This locks in profit whilst leaving funds to ride out temporary dips and extract gains from extended JSE runs.

Note that BITS-45, BITS-41, BITS-MT and BITS-ST signals, alerts and charts are provided as part of ENHANCED/PRO subscriptions only. If your contract is older than 3 months, then it is a STANDARD contract and you need to email us if you want your contract upgraded to PRO. To see the differences between a STANDARD and PRO contract, go HERE.

4. HISTORICAL PERFORMANCE AND SIGNAL CHARTS
Using the 4 BITS variations in a JSE investing strategy coupled with a fixed interest vehicle such as a money market fund or fixed deposit account earning prime less 4% during non-vested periods produces the total return curves shown below (brokerage of 0.75% is assumed and income taxes and dividends excluded):



We can see that BITS41 is having the benefit of staying vested in the more recent past whilst BITS45 is stuck in cash. Recent vested periods of the two strategies since 5th January 2009 are compared below with the vested periods represented by the brown vertical shaded areas. It is interesting to note that all the BITS strategies entered the market on 23 March 2009 some 20 days after the absolute bottom on 3rd March 2009.



A longer term historical view of the various vested periods since March 1996 appears below. The BUY threshold of 60 is shown with the horizontal green line and the "SELL" signal by the red line. You can click on the first chart for a larger, higher resolution picture.









The one thing we can observe is that the BITS strategies certainly do not latch onto ALL the JSE's strong runs and in fact miss out on some pretty good JSE gains at times. But when it does latch, the latch is good and has high confidence. For this reason it is advisable to use BITS in parallel with other strategies such as SwissClock and BPI to try and capture as much JSE gains as possible and to provide for strategy diversification as laid out in "The PowerStocks Way".

More importantly, note how during the August 1997, June 1998, February 2000 and brutal 2002-2003, March 2004, Feb 2005, July 2007, Nov 2007, and bone crushing May 2008 bear markets that the BITS strategies AVOIDED the JSE rather nicely. That is a lot of bear markets to be neatly side stepping to be put down to mere chance, and instrumental in the strategies' phenomenal win rates.

RELATED TOPICS --> | Large resolution historical BITS Signal chart |
                             | The Martin Zweig BITS Signal variations |
 
Make a Free Website with Yola.