Determining the Optimal Market to Trade

DETERMINING THE OPTIMAL MARKET TO TRADE

The three bearish engulfing patterns I just outlined all occurred on the same date and yet provided different results for short positions. Like I’ve said before, trading is a statistical business and not every pattern results in a profitable trade. The trader must understand how often the pattern produces a profitable result and the effects of various parameters such as volume, price level, and average volume, as well as the effects of varying different parameters involved in the definition of the pattern. Bottom line, it’s time to start testing again.

FIRST TEST PERIOD: 1/3/06 to 5/1/ 07

Figure 3.4 shows the results of testing the bearish engulfing pattern during the period of 01/03/06 to 05/01/07. The results show that during this period the pattern not only performed worse than buy and hold, it actually lost money. In fact, losing trades occurred more than 55% of the time and less than 44% of the trades during this period resulted in a profit. The initial backtesting results raise a caution flag.

FIGURE 3.4: BEARISH ENGULFING BACKTEST DURING 1/03/06 TO 5/01/07

TEST PERIOD INCREASED TO ONE YEAR

The market is always changing, and during certain market conditions some systems perform better than others. It is important to check a potential trading system in multiple time frames and also in different market conditions. Figure 3.5 shows the results of increasing the test period by a year and testing the basic bearish engulfing pattern during the period of 01/03/05 to 05/01/07.

FIGURE 3.5: BEARISH ENGULFING BACKTEST DURING 1/03/05 TO 5/01/07

Increasing the length of the testing period by one year significantly changed the results. The bearish engulfing pattern still shows a loss for the test period, but it was reduced from a loss of 35% for the first test to a smaller loss of 26% when adding a year to the test period. One possible explanation for this is that the results of using the bearish engulfing pattern depend on the market conditions in which it is used.

In order to test this theory, we have to consider what different types of trading environments the market can present us with. We know there are only three things the market can do. It can move down, move up, or move mostly sideways. Testing the bearish engulfing pattern in each of these three different market conditions will let us know if the pattern works better in one of the three market types.

DOWNTREND MARKET

During the period of 12/31/04 to 04/28/05, the market was in a downtrend and dropped about 200 points, or about 10%, as shown in Figure 3.6. Testing the basic bearish engulfing pattern with a three day holding period during this period resulted in 1,859 trades and a 31% annualized ROI, which is quite favorable as compared to the approximate 10% loss in the NASDAQ. Profitable trades equaled about 58% and only 40% resulted in losses. This is a significant improvement over the previous results.

FIGURE 3.6: DECLINING MARKET DURING FIRST FOUR MONTHS OF 2005

SIDEWAYS MARKET

During the period of 11/17/05 to 03/13/06, the market moved mostly sideways within a 100 point range, as shown in Figure 3.7. Testing the basic bearish engulfing pattern during this period, when the market was stuck in a trading range, resulted in 2,880 trades and a 45% loss. Fewer than 45% of the trades were winners and more than 54% were losers. It would appear that at least in these two cases, the basic bearish engulfing pattern performs better when the market is in a downtrend than when it is moving sideways in a range. 

FIGURE 3.7: TRADING RANGE MARKET DURING DECEMBER TO MARCH 2006

UPTREND MARKET    

The third market condition is an uptrend, such as the market experienced during the period of 08/11/06 to 11/17/06, as shown in Figure 3.8. Running the basic bearish engulfing pattern with a three day holding period during this bullish market period resulted in 2,627 trades that showed an annualized ROI of negative 69%. Less than 37% of the trades were profitable and the trading pattern lost money on more than 62% of the trades. This was the worst result so far.

FIGURE 3.8: BULLISH MARKET DURING AUGUST TO NOVEMBER 2006

The test results are interesting and show that in these cases the bearish engulfing pattern works best in a downtrending market and should be avoided in sideways or uptrending markets. Testing the basic bearish engulfing pattern in another market downtrend, the period between 07/28/05 and 10/17/05, confirms the previous results by showing a 38% annualized ROI for the 1,451 trades made. More than 54% of the trades were profitable.

After testing the bearish engulfing trading pattern in two different time frames and all three types of market conditions, we find that the only times it has shown profitable annualized ROIs were the two times it was tested in bearish markets. Additional testing bears out the results that this trading pattern works best in declining markets and provides poor results in trading range or bullish markets.

Many trading patterns show specific market environments in which they perform well; and others where they are best avoided. One of the keys to trading involves knowing which is which. Unless traders study the results of large numbers of trades in different market conditions, they will not know when to use each pattern in their trading toolbox. Using the same pattern in all market conditions may just churn the account. Selecting the right pattern for the current market conditions may improve trading results.

VARYING THE HOLDING PERIOD

Table 3.1 shows the effect of varying the holding period for the basic bearish engulfing pattern when running the test during the market downtrend of 12/31/04 to 04/28/05. The best results are obtained using a four day holding period and the annualized ROI drops off consistently for longer holding times. The bearish engulfing pattern is a short term trading pattern and the longer the position is held, the lower the results. Each trading pattern seems to have time frames that it likes best. It is a good idea to check this information for every pattern you intend to use.

TABLE 3.1 TEST RESULTS FOR 12/31/04 TO 04/28/05 WITH DIFFERENT HOLDING PERIODS