Trend following techniques are the most popular among mechanical traders. While these methods do work, they have a given footprint. Many trend following systems have more losing trades than winners with potentially longer drawdowns. Countertrend systems have a higher winning percentage but a smaller average winning trade size.
Countertrend systems buy weakness and sell strength. Just like trend following systems, they offer a special set of design issues. Normally, for countertrend systems to be successful, they need a winning percentage above 50%, preferably above 60%. This means win/loss ratios are lower than 2-1. Although it is difficult to quantify why, countertrend systems seem more dependent on individual market characteristics and are more often designed for a given market or asset class.
Trader Larry Connors developed a simple countertrend system that he popularized in his book “Short-Term Trading Strategies That Work.” This system is called 7-7 because it buys and sells on a seven-day low or seven-day high.
It also is designed to deal with a flaw of countertrend systems. Because these strategies buy weakness and sell strength, they can be badly positioned in a market crash or “irrational exuberance.” Connors added an average close filter to combat this. The rules are as follows:
If Close=Lowest(Close,BuyLen,0) And
Close>Average(Close,FilterLen,0) Then Buy(“Buy-7”,1,0,Market,Day)
If MarketPosition=1 and CLose=highest(Close,ExitLen,0) Then
The standard parameters are 200, 7 and 7, but by designing them as inputs we can test other combinations.
Connors designed this system to trade major index exchange-traded funds (ETF), such as SPY, QQQQ, FXI and EWZ. In an effort to understand countertrend systems, we will test this system on a basket of futures markets and explain why it works on some markets but not on others. The results in “Split sevens” (below) are from Oct. 7, 1997, to Aug. 6, 2010. (The original system was long only because it was built for ETFs . We added a short component.) The combined results for both markets are shown in “Seven summary” (below).
Optimizing this system reveals that shorter sets of parameters make more money but could be more sensitive to slippage and commissions. Also, longer sets of parameters don’t make as much money but have larger average trades. For example 200, 10, 10 make $956,904.38 on the long and short side combined with an average trade of $435.15.
The combination 125, 2, 2 makes the most money when we don’t deduct slippage and commissions but the average trade is low. We have a total profit of $1,593,629.06 but only $98.09 a trade on 16,247 trades.
If we take a closer look we will see that some markets did well on long but not short. One example of this is copper. Copper made $52,000 on the long side but lost money on the short side. Some markets like cotton just did not work at all with this system.