In this post I analyze the “Double 7s Strategy”, originally revealed in the book Short Term Strategies that Work: A Quantified Guide to Trading Stocks and ETFs, by L. Connors and C. Alvarez. The backtesting results indicate that this popular and supposedly simple strategy has deteriorated significantly since publicized. I list a few possible explanations as to why the strategy has deteriorated at the end of the post.
Derek Hernquist’s blog, Musings of a Tape Reader, is a source of interesting, thought provoking ideas about trading, systems and the markets in general. In his post Market Noise and Signal, on November 3, 2012, (link at the end) Derek pointed to a paper by Nick Libertini of 361 Capital. The paper started on interesting exchange of ideas in the comment section and another link post by Derek about the “Double 7s Strategy”. I will make another post about the results in the paper by Libertini. In this post, I would like to concentrate on the “Double 7s Strategy” because of its popularity.
In a special report issued by the authors of the aforementioned book, the strategy details were given as follows:
The SPY is above its 200-day moving average
If the SPY closes at a 7-day low, buy.
If the SPY closes at a 7-day high, sell your long position.
The above strategy involves only three rules and it is easily expressed in natural language but, fundamentally, this is not a simple strategy because it involves 3 free parameters.
The special report that revealed the above strategy involved backtesting results for 4 ETFs: SPY, QQQ, FXI and EWZ. All showed exceptional performance since inception. One comment I would like to make here is that the overall sample size was very small, only a total of 310 trades for all 4 ETFs and for FXI only 26, but I will not elaborate on this issue here.
Back-testing the “Double 7s Strategy”
I decided to backtest the strategy first using SPY daily data to try to identify potential sources of weakness. Below is an equity chart of the backtest starting at inception of this ETF to 11/02/2012:
I immediately spotted the first potential source of weakness of this strategy from the above chart. I have placed a blue vertical mark at the point where one can see that the bulk of the returns of this strategy was made during the smooth uptrend that resulted from the bull market of the 1990s. Look at the chart above and notice the parabolic rise in equity that came to an abrupt end when price moved below the 200-day moving average. After that, the equity curve moved almost linearly with high volatility and long periods of flat performance because no trades were triggered, like during the 2001 -2002 and 2008 bear markets.
Table 1. “Double 7s Strategy” SPY performance for two distinct time periods
Table 1 shows the performance for the two periods marked on the chart based on $20,000 starting capital and a fully invested system (No commission or slippage included). It is clear that the strategy deteriorates fast after 09/2000 and this can be seen especially from the low payoff ratio (ratio of average win to average loss).
However, the strategy is still profitable in SPY, although we have identified a potential source of weakness, i.e. the fact that the strategy performs well during smooth uptrends and, as a result, the performance shown in the special report where the strategy was disclosed was skewed severely by the parabolic rise in equity that ended in 09/2000.
Forward testing the strategy for the other ETFs will provide further indication of its robustness. First we start with FXI that in the special report showed only 26 trades. Below is the equity chart since inception to 12/31/2007, i.e. the period used in the special report:
The performance is positive but another source of weakness may be identified here. Although price stayed above the 200-day moving average after August 2007, performance was flat and I attribute this to the fact that a downtrend started in October of that year. Thus, this is another indication that for this strategy to work a smooth uptrend is required.
Below is a chart of the equity curve for the forward period after 12/31/2007:
The performance here is obviously disappointing. Again we see that the equity curve rises when there is a smooth uptrend but as soon as prices move sideways, a drawdown occurs that wipes out the gains. Below is a table of the performance for FXI during the original and forward periods:
Table 2. FXI performance
The deterioration in performance is evident in the forward period with the net profit barely staying in positive territory.
Now, let us turn to EWZ. We repeat the same study for the original backtest period and forward period. Below is the equity chart for the backtest period:
Here we also see a parabolic rise in equity. Note however that for the most part, price has stayed above the 200-day moving average and the market was on a strong uptrend.
Below are the results for the forward period, after the backtest period in the special report:
Here we get negative performance that is primarily due to the fact that price moves either below the 200-day moving average or sideways and there was an absence of a smooth uptrend. Table 3 shows the performance metrics:
Table 3. EWZ performance
Lastly, let us look at the fourth ETF in the special report, QQQ. Below is the chart of the backtest period performance:
As expected a positive performance but with extended periods of no activity due to price falling below the 200-day average. From the above chart we also notice the main problem of the strategy, specifically that it is negatively affected when price fluctuates around the 200-day moving average and to perform well it really needs a good smooth uptrend.
Next is the forward performance chart. Here we notice an immediate drawdown and the overall performance is negative:
Table 4 summarizes the performance metrics:
Table 4. QQQ performance.
This one was basically a disaster. Again, for the same reasons stated above.
Below is a table that summarizes the profit factor performance of the 4 ETFs for the in-sample and forward periods:
|ETF||inception – 12/31/2007||01/02/2008-11/02/2012|
Recall that, to have net profit > 0, the profit factor must be greater than 1. Therefore, the strategy failed in QQQ and EWZ and was marginally profitable in FXI.
It appears that the “Double 7s Strategy” works well only if there is a smooth uptrend but without extreme corrections. This implies that the strategy may be curve-fitted. This hypothesis is strengthened by the fact that the strategy failed forward tests in 2 out of the 4 ETFs and marginally survived in a third. However, there are some other possible reasons as to why the strategy degraded. Below, I list a few to stimulate thinking:
Reasons for the strategy degradation
1. The strategy was originally curve-fitted.
2. The strategy was the outcome of an adaptation process, similar to genetic evolution and as such subject to Fisher’s law. I will talk about this in more detail in another post.
3. Reflexivity (Soros). The strategy became very popular after being published and actually caused an increase in volatility just enough to degrade it because strong hands faded the retail traders using it.
4. The strategy was random, an outcome of selection bias.
I will go with number 3, if I may. I think this was a good strategy that was simple to trade even with daily quotes off a newspaper and a calculator. Thus, many retail traders used it and as a result it lost its edge. Maybe its authors set up an experiment by revealing it but I do not want to speculate further on this one. Obviously, they are very smart individuals.
Related links and posts:
Disclosure: no relevant position at the time of this post and no plans to initiate any positions within the next 72 hours..
Charting program: Amibroker (Charts created with AmiBroker – advanced charting and technical analysis software. http://www.amibroker.com/”)