A revolutionary concept developed by Greg Morris in 1991, called candle pattern filtering, provides a simple method to improve the overall reliability of candle patterns. While the short term trend of the market must be identified before a candle pattern can exist, determination of overbought and oversold markets using tradi­tional technical analysis will enhance a candle pattern's predictive ability. Concurrently, this technique helps eliminate bad or pre­mature candle patterns.
One must first grasp how a traditional technical indicator responds to price data. In this example, Stochastics %D will be used. The stochastic indicator oscillates between 0 and 100, with 20 being oversold and 80 being overbought. The primary inter­pretation for this indicator is when %D rises above 80 and then falls below 80, a sell signal has been generated. Similarly, when it drops below 20 and then rises above 20, a buy signal is given. (See Chapter 10 for more on Stochastics.)Here is what we know about stochastics %D: When it enters the area above 80 or below 20, it will eventually generate a signal. In other words, it is just a matter of time until a signal is given. The area above 80 and below 20 is called the presignal area and represents the area that %D must get to before it can give a trading signal of its own The filtered candle pattern concept uses this presignal area. Candle patterns are considered only when %D is in its presignal area. If a candle pattern occurs when stochastics %D is at, say 65, the pattern is ignored. Also, only reversal candle patterns are con­sidered using this concept.
Candle pattern filtering is not limited to using stochastics %D. Any technical oscillator that you might normally use for analysis can be used to filter candle patterns. Wilder's RSI, Lambert's CCI, and Williams' %R are a few that will work equally as well. (These oscillators are explained in Chapter 10.)