A trader’s arsenal includes many weapons. Some are extremely useful, and some have limited utility. A successful trader will understand the strengths and weaknesses of many techniques to combine various tools so the combination proves greater than the individual parts.
Success in the trading business may be defined as simultaneously increasing net profits while reducing overall risk. So, if trading tool A produces net profits of 240 points and has a winning ratio of 61%, and trading tool B produces net profits of 150 points and has a winning ratio of 72%, then a new tool AB certainly will be quite successful if it can produce net profit of 275 points with a winning ratio of 74%.
We even might consider the new trading tool successful if it does one or the other — increase net profits or increase winning percentage. The assessment depends on many factors, including individual trading personality, goals and risk tolerance. Nevertheless, we always seek to improve on our tools in search of perfection.
Methods at hand
The two-part series, "Using price shocks to time the market" (Futures, November 2009) and "Shocking profits in new markets" (Futures, December 2009), discussed the concept of entering a market in the opposite direction when an extreme price level, in the context of the most recent price action, had been reached.
We examined various commodities and compared monthly closing prices to those occurring during the previous two years. It was hypothesized that a trader could enter such markets safely when those markets exhibited signs of momentum deterioration and then hold the position long-term until the pendulum swung in the opposite direction. This tool is quite useful, but the challenge is quantifying the level of extreme that a market can reach.
As the articles pointed out, analysis demonstrates a market has a natural floor because commodities have real production costs. However, as demonstrated by the recent price increase in cotton exceeding 550% over the November 2008 low of 36.70¢, sometimes the sky is the limit on the upside.
In "S&P 500: Moving averages provide a simple solution" (Futures, April 2011), a useful improvement on a basic tool was developed. By shifting a simple 15-period moving average forward by 14 periods, the trading results improved dramatically, both in terms of profitability and accuracy. In addition, the refinement reduced overall risk exposure while retaining the best trades realized using the simple moving average tool.
The challenge before us is to combine both concepts to determine if we can realize an improvement.
To conduct our test, we chose a market at random — copper on a monthly basis. Our data set begins in January 1989 and ends in March 2011, providing 267 data points. To perform the test, we shall begin by using the methodology developed in the displaced moving averages article.
That is, we buy copper when the second monthly close exceeds the shifted moving average. We sell any long positions and go short when the second monthly consecutive close is beneath the moving average. This will be our control system.
As demonstrated in "Control system" (below), the results were not satisfactory for this system by itself. A total of 10 trades were taken over the 21-year period, including one open trade at the time of this writing. Of those, only 40% were winners, and the overall loss before slippage trading one contract is 16.7¢, or a gross loss of $4,175. The average trade lost $417.50 before slippage and commissions. Because the last trade is still open, chances are good these results will worsen.
To address these poor results, we will introduce the concept of a profit target suggested in the moving averages article and, borrowing from the price shock series, an opposite side entry if it is determined that price has exceeded its historical price pattern in either direction.
This is where discretion comes into play. These levels are based on observation of recent price action and reading the patterns. If testing demonstrates that 95% of the data fall within a certain range, by definition, 5% will fall outside with reasonable statistical probability and less chance of error. Naturally, the exact range is an estimate, not a certainty, and its accuracy will depend upon the number of data points available to formulate the estimate.
For this test, chart analysis suggests we use approximately 25% of the range as our profit target and approximately 50% of the range as our threshold for an extreme price move.
We will buy copper as before, but exit the long position when it has achieved a price equal to 25% above the displaced simple moving average. Similarly, we exit shorts when they drop to a price 25% beneath the displaced average. This is triggered on a market-if-touched basis, and these levels change monthly in relation to the movement of the displaced average itself.
We integrate the price shock concept by entering an opposite direction trade based on the 50% level. So, if price closes at least two months beyond the 50% upper level and then closes beneath that level in a subsequent month, we enter a short trade. We hold that short and look to exit at the profit target.
This approach now combines several simple trading tools: 1) a simple moving average; 2) displaced by roughly its own period; 3) with entry based on a filter, in this case consecutive months; 4) a profit target and 5) an opposite trade using the shock methodology; but 6) only after a filter has been overcome. All six of these are computed easily and a trading determination can be made. There can be no question about any trade entry or exit with the sole exception of what happens if price exactly equals a filter level, which is so unlikely that it’s trivial for the purposes of this test.
The results are considerably better, as demonstrated in "Final system results" (below). The win/loss ratio over the same 10 trades has doubled to 80%, which is phenomenal. The total profit is $4.0814 (vs. the 16.7¢ loss for the control system), or a total gross profit of $102,035, and all trades are closed. The average trade gained $10,203.50 before slippage and commissions for each contract traded. Trading time ranged from 59 to 943 days and the average trade lasted 268 days; 70% of the trades lasted roughly one year or less.
Another important result is the reduction of realized loss. The control method suffered four relatively minor losses, but also took an enormous hit of 172.10 points. The improved method had only two small losses.
Nevertheless, it is important to remember the methods both can suffer significant drawdowns before closing trades, even if the result of such trades is profitable. For example, the October 2007 short trade entry triggered by the shock methodology at 347.30 produces a nice profit of 96.43 gross points, but a trader would have had to hold through a drawdown of 79.70 points in May 2008 prior to realizing the eventual gain five months later.
Such drawdowns are more likely in the shock trades because they result from extremely strong momentum creating the situation. Although such momentum is deteriorating, traders often test or exceed the earlier levels even though the strength of such moves is fading, as we observed in "Trading stochastics in context" (Futures, June 2011). The advanced trader could add the stochastics divergence filter as an additional tool for timing market entry in the shock situation.
Because copper exceeded the 50% shock level in October 2010, and remains there as this article is written, it is anticipated at some point an opposite side shock trade will be set up some time during summer 2011. Were this to occur on schedule, a drop from the $4.25 area to the $3.15 area is likely with reasonably high probability. A potential $1.10 drop in copper sets up a $27,500 potential per contract profit before slippage and commissions. Because copper has an initial margin of roughly $6,000, a medium-size account could easily put on between one and three contracts and seek to return roughly 450% of margin allocation — an excellent return on investment.
Additional tests should be run on other commodities to validate the methodology and determine precise profit and shock levels. Naturally, each will have a unique pattern and levels for profit and shock, based upon the nature of the underlying commodity market and the recent price action in out-of-sample data. As always, a short-term trader could employ similar methods on 60-minute bars. The concepts herein are robust and demonstrate how combining different simple tools can result in a far superior approach than the core methodology on which they are built.
Arthur Field has a Ph.D. from Clemson and is a former fund manager for Fidelity International. He wrote "The Magic 8: The Only 8 Indicators You Need to Make Millions in the Markets," available at www.themagic8.com. Email him at email@example.com.