The rate of change of an object will show decreasing momentum until the object actually changes direction.
The big criticism of oscillators is that they sometimes give a trading signal but the market is in such a strong trending mode that the signal is false. The common wisdom is that oscillators do well in nontrending markets
and poorly in trending markets.
The simpler the oscillator, the more sensitive it is to current market price action. For example, a simple oscillator based on the 10-day rate of change will be more sensitive to current price action than an oscillator
based on the 30-day rate of change.
Many analysts got beat up badly using the simple oscillators so they tried to find ways to improve them. Welles Wilder’s relative strength index (RSI) and stochastics are the two most popular and famous improvements on the basic oscillator. Larry Williams’s %R is actually the same as stochastics except not smoothed and the scale is turned upside down. Stochastics are reasonably easy to calculate. There is a three-step process to calculate the two components of stochastics: %K and %D.
First calculate the raw stochastic number:
K = (C − L)/(H − L) ∗ 100
Where C is the close of the last x number of days, H is the high for the last x number of days, and L is the low for the last x number of days. Typically, x is the last 14 days, though 9 days is also very popular.
%K is a three-day MA of K, and %D is a three-day MA of %K
Fortunately, you won’t need to know the math unless you want to fiddle with it to see if you can improve it. Virtually all chart services, quote systems, and software packages have stochastics built in. However, they
sometimes fiddle with the parameters. I suggest using the parameters I just outlined. The software packages and quote systems often allow you to change the parameters.
The one important question that must be addressed is the term of the stochastics. How many days will you use in your calculation? The classic answer is 14 days, which many people claim is about half of the normal
cycle of forex prices. Other people use nine days, though I have never heard a justification for this number except that it is more sensitive than the usual 14 days. I have never seen any proof of the claim that the average price cycle is 28 days. Further, I have never seen any argument that you should, therefore, use a half cycle as your length in using stochastics. Nonetheless, fewer days in your calculations will lead to a more sensitive stochastic.
Could the length of stochastics be optimized? (By this I mean testing different parameters to find the most profitable one over a given test over past data.) Yes, but first you have to determine how you are going to use
them. Are you going to use them as an overbought/oversold indicator? Are you going to use them as a trend indicator? Each use of stochastics will likely lead to a different optimization.
I won’t get into the pros and cons of optimization here, so I suggest you read Bob Pardo’s book on trading systems. See the Appendix for more information.
Unless otherwise stated, I will use 14 days as the length of the stochastics. Why? Well, I came up with all these profitable techniques when everything had to be done by hand and 14-day stochastics were all we had to use when testing. In the meantime, I have seen that these techniques work so I am loath to change them. I leave it to my superintelligent readers to optimize them!
A second reason to use 14 days for the length is that readers can easily use the indicators because they are the most common in available software packages. There is no need to fiddle with your software package to figure out how to change the parameters.
The techniques outlined in this chapter work similarly on both 9- and 14-day time frames. This is an important suggestion that stochastics are a robust indicator. However, the main difference between the 9- and 14-day
stochastics is that the 9-day is more sensitive and gives more indications or signals. A key consideration in determining which length to use is your particular trading style. A longer length will give fewer signals and later
signals but will filter out some whipsaws. A shorter length will give more signals and earlier signals but will be whipsawed more often.
Stochastics are essentially a measure of the close as related to the high and low. The stochastic measures the percent distance of the close to the range. Thus, a reading of 50 percent means that the close is halfway between the high and low. A reading of 75 percent means that the close was at the 75 percent level between the high and low. In other words, it was at the 75 percent level of the day’s range or nearer to the high than the low.
This means that you can only see a stochastic reading of 100 percent if a market closes on its high every day in the study. The underlying concept is that it is bullish if the market is tending to close in the upper half of the
day’s range and bearish if the converse is true.
Stochastics are used to trade all time frames, including day trading. Obviously, the most popular time frame is daily, using daily bars. However, I know a number of people who use them on shorter-term bars, such as
five-minute and 60-minute bars. George Lane, the inventor of stochastics, used to use stochastics on three-minute bars in the S&P 500 futures.
Source: How to Make a Living Trading Foreign Exchange: A Guaranteed Income for Life (Wiley Trading)
