Current Monthly Trading Article

Buying Quality Stocks in Today's Market: Pullbacks Versus 21-Day New Highs and 21-Day New Lows

I’ve written here about cycles in the market, i.e., market patterns that appear often enough to be judged statistically significant. For example, the Automotive, Health Services, and Aerospace Industry Groups were shown last month to provide the best performance of the 31 groups in the September-to-December time frame over the last four years. Here, I want to contrast three chart patterns that are routinely used to enter long positions in short-term trades. I think you’ll be surprised at the results.

Larry Connors and Conor Sen wrote an important book detailing the importance of chart patterns, “How Markets Really Work: A Quantitative Guide to Stock Market Behavior”. I received my copy just before hurricane Ivan arrived and was so fascinated by the subject that I finished it that evening, storm and all. It presents 15 years of historical data (1/1/89 through 12/31/03) refuting much of what passes for "common market truisms." These authors looked at several market criteria (new 5 and 10-day highs and lows, the impact of being above or below the 200-day moving average, multiple day pullbacks, and others) to evaluate how each impacted one day through one week returns for the S&P 500 and Nasdaq 100 cash markets. Most of what they found went against our dearly held market beliefs.

For example, during this period there were 947 10-day highs in the S&P. Buying the close of a new 10-day high then selling one week later returned 0.0% on average with 53.43% of the trades being profitable. Not too impressive, since the average 5-day return (irrespective of whether a new 10-day high had been made or not) was +0.21%. Contrast this behavior with buying a 10-day low then selling one week later. This strategy returned 0.56% on average with 333 (of 557) profitable trades (almost three times the unrestricted average 5-day return). The message: Buying strength hasn't worked as well as buying weakness! If one buys the 10-day low but requires the S&P 500 cash index to be above its 200-day moving average, the average return increases to +0.66% with a 66.43% win rate (190 winners of 286 occurrences). The authors put it this way: "...the greater opportunity and edge lies in being a buyer as the market makes a new short-term low versus buying when it makes a new high."

Following this theme, I tested three common patterns used for entering long positions in individual stocks: three or more day pullbacks, new 21-day highs (breakouts), and new 21-day lows (bottom fishing). Chart 1 shows examples of each for ARO as they occurred over a two-month period in 2003. To validate this study, I studied these three patterns for a group of fundamentally sound stocks (9/24/04) with earnings and analysts’ rankings revision fuel (Zacks rankings of 1 or 2) and value left in price (two year PEG ratios less than 1.25): 36 stocks shown in Table I were evaluated over the 294-day period between 7/30/03 and 9/28/04 (10,584 test days).

For the pullback, three or more lower highs were followed by a reversal candle. To limit purchases to stocks in up trends, the 50-day moving average was required to be rising over the last six days, and the 20-day moving average to be greater than its 50-day. When these conditions were met, a long position was entered then exited at the close five days later.

Similarly, for the 21-day high, a position was entered when today’s price exceeded the high of the last 20 days, again, when the 50-day moving average was rising over the last six days, and the 20-day moving average had to be greater than its 50. When these conditions were met, a long position was entered then exited at the close five days later.

Finally, for the 21-day low, a new position was taken today when price exceeded the high of yesterday’s new 21-day low, this time with no additional conditions. And again, the position was exited at the close five days later.

Table 1

Table I provides results for each of the 36 stocks over the 294-day period. AMED, for example, had 16 pullbacks meeting the required conditions and an average resultant five-day gain of +3.0% (10 with positive gains and 6 with losses); AMED had 33 instances of making new 21-day highs and an average five-day gain of +3.89%. AMED also had 6 instances of new 21-day lows and an average gain of +5.58%. Contrast these returns with the average 5-day gain (+2.29%) earned from the control condition, i.e., from buying each occurrence where today’s price exceeded yesterday’s high and holding for five days.

For these stocks, the average 5-day gain for the control condition was +1.27%, as these were quality stocks performing in both bullish and bearish phases of the market. Limiting buys to pullbacks increased returns to +1.84% (1.45x the control), to new 21-day highs increased returns to +1.38% (1.09x the control), to new 21-day lows increased returns to +2.59% (2.04x the control). Clearly, in this market, buying the new 21-day lows for a week-long trade was a superior strategy to either buying pullbacks or buying breakouts. These results are consistent with Connor’s findings for 15 years of S&P data. While significantly longer holding periods lessens the importance of an entry strategy, the strategy of buying 21-day lows reduces short-term risk and provides better risk/reward limits.