Taking signals

Posted by Scriptaty | 8:29 PM

However, there is a problem with using a 14-period Donchian channel and a 14-period stochastic. Identical lookback periods do not make sense if you’re trying to determine both the long-term trend and shorter-term overbought and oversold points to signal entry points.

Trial and error led to adopting a 70- period Donchian channel as the trend filter, with the simple rule that if the two channel lines are stretching apart, the market is trending. If the channel lines are contracting, the market is in a trading range.

When the market is in a trading range, the stochastic can be used to buy oversold points and sell overbought points. When the market is trending, take only those stochastic signals in the direction of the trend.

It shows the same price action, with 70-bar Donchian channel lines added. Beginning in December 2000 the difference between the channel lines leveled off and the market entered into a trading range. In February 2001 the two lines contracted (the lower line was static while the upper line declined) and the stochastic indicator prematurely signaled a bottom in March. However, the indicator gave another crossover buy signal in June, just ahead of the July lows. There was a crossover sell signal just before the September 2001 peak.

During June 2002, price began to push the upper Donchian channel line higher while the lower channel line moved sideways. This stretch signaled that the market was moving into an uptrend. In this situation, you would look for the stochastic to pull back and give a buy signal. Because the market is
trending it is not necessary for the stochastic to fall all the way to 20 or lower; if it does, however, it is still a valid buy signal if the channel lines are stretching. For example, in October and December 2002 and April 2003, %K crossed above %D at opportune moments to enter the bull trend during pullbacks. In September 2003 and April 2004, the stochastic did, in fact, drop below 20 when generating %K/%D crossover buy signals.

0 comments