by Lee Gettess

Stochastics, Relative Strength Index (RSI), Moving Average Oscillators… virtually every technical trader attempts to make use of at least one of these tools, but the statistics show that very few of them meet with success.  Does that really mean they are worthless tools?  No, not if they are employed properly.

Let’s get the basics out of the way. My preference is to use a 5-period RSI, a 9-period Slow Stochastic, and a Moving Average Oscillator made up of the difference between a 5-period simple moving average of (H+L+C)/3 and a 22-period simple moving average of (H+L+C)/3.

The complete formulas for RSI and Stochastics are somewhat intricate.  Both are normally part of any commercially available graphics package, and it is certainly much easier to just let the software calculate them for you.

The Moving Average Oscillator is pretty straight forward.  For each price bar you add the high, low and close together then divide by three.  That is the figure we use for each price bar.  That gives you the “fast” moving average.  You also add the last 22 of them together and divide by 22, which gives you the “slow” moving average.  You then subtract the value of the slow from the fast.  The result is the actual oscillator.

I use the particular values that I do for these indicators for what, at least to my warped mind, are logical reasons.  Most traders think in terms of days, weeks and months.  The 5-day RSI tracks what amounts to the weekly cycle.  The 9-day Slow Stochastic is just a little bit quicker than the standard two week cycle, which would be 10.  Then, since we are quicker on the two-week period, the moving average oscillator tracks the 1 week compared to the 4 week, only we go a little slower.  The 5 moving average is equal to 1 week, then the 22 moving average is a little slower than 4 weeks, which would be 20.  All of this has its basis on daily charts, but the relative relationship remains intact regardless of what time frame you choose to operate in.  It is just as valid on a 3 minute chart as it is on a monthly chart.

Now, are these indicator values critical for the success of this methodology?  I doubt it.  I wouldn’t’ even claim that they are the parameters that will give you the absolute best results.  They are simply the values that I chose to work with originally, and they seem to show me what I want to see.  Feel free to experiment with other parameter settings if you like.  I will consistently use these, and I believe consistency is the most important factor.

Both Stochastics and RSI have relative values that are generally considered to be overbought or oversold.  These are often 20, 25, or 30 for an oversold condition, and 70, 75, or 80 to be overbought.  While those readings can be monitored to get a clue as to the overall condition of the market, there is no absolute reading that is required to make use of this methodology.

What we are looking for from these three indicators is divergence.  That is, when price makes a higher swing high than the previous swing, but the indicator has a lower reading.  That is a divergent high, which sets up a potential sell.

For a buy set up, you want to see price make a lower swing low while the indicator has a higher reading.  We can quibble about whether an equal high or low, or even an equal indicator value can be considered divergence… it isn’t terribly important.  My advice would be to always assume you do not have a signal unless it is obvious that you do.  I sometimes do accept equal readings, but normally there is other evidence to make me come to that conclusion.

Now, divergence on any single one of these three indicators may be an excellent signal for an impending market turn.  HOWEVER, IN ORDER FOR US TO BE INTERESTED IN TRADING THIS SIGNAL, WE NEED TO SEE DIVERGENCE IN AT LEAST TWO OUT OF THE THREE.  Often all three will be divergent, which is a great confidence boost since it indicates a lessening of momentum on three distinct time frames, but two out of three is sufficient for us to look to take a trade.  Whether one reading is higher or lower than another is not always evident to the naked eye, but most all graphical software will allow you to check the exact mathematical value so you can be sure.

Anyone who has worked with divergence in indicators previously knows what a good signal it can be… and how bad!  In an especially strong trending market you will see divergence, followed by divergence, followed by divergence, etc… meanwhile the market continues on without changing direction and your account continues to dwindle from bucking the trend.  We will attempt to filter our erroneous divergence signals with the use of the price pattern.  Divergence by itself, even with all three indicators, is simply not enough if price doesn’t agree.  We need a pattern that will give us reliable entries before the market runs too far off of the divergent high or low that it is making.  We want to get positioned near the high or low, but not until we receive price pattern confirmation that we are most likely making a high or low.  Fortunately, we have a pretty decent pattern to use.

Back in 1987, (I sound like a very old man, don’t I?) I worked with a trader named Joe Duffy on a price pattern that seemed to indicate reversals in the market.  Joe used it exclusively on daily charts, and we designed a short term trading system called Pulsar which used this pattern as its basis.  While it did work well for short term trading off of daily charts, my research had shown it setting up on virtually any time frame.  In many cases, the moves off of the pattern were quite significant rather than just quick pops.  Sometimes it would occur at the low of the day on a 5-minute chart, and sometimes a 30-minute chart would indicate the low for the entire week.  It worked like gangbusters!  The problem was attempting to isolate the situation where it would lead to a longer term move, rather than just a quickie scalp trade.  Although I am sure you would agree there is nothing wrong with a quickie!  At any rate, I kept watching this pattern develop on various time frames and trying to identify the situations where it worked best.  Eventually things started coming together by adding our previously discussed indicators.

Here is how the pattern sets up: For a “buy” situation the market must come down and make a low of some degree, then have a higher close.  This higher close can occur on the same day as the low, but can also occur on any ensuing day.  Incidentally, I am using the term “day” for discussion, but “bar” would be more correct since this is applicable to any time frame chart.  After the higher, or up close, the market must close lower without taking out the previous low.  That is, a day must close lower than the previous day’s close, but not trade lower than the lowest low that we are working off of.  There may be any number of up closes in a row before the down close.  All we care about is the day of the first down close.  The high of the day of the first down close becomes your breakout point, and you would place a buy stop one tick over the high of that bar.  Your protective stop would be placed one tick below the lowest low that began the pattern.  There may be several down closes in a row, or any combination of up closes and down closes before we actually enter the market.  The only thing we are concerned about is the first down close day, and our buy remains one tick over the high of that day, provided the market does not make a new low in the interim.

Basically this pattern is a low, followed by an up close, followed by a down close.  We buy one tick over the high of the down close day with our stop loss one tick below the original low.  This pattern can occur in just two days, or it may take any number of days to form.  Originally, the requirement was for the low to be of some degree… for instance, it had to be the lowest low of the last eight days.  Rather than that, we are going to require that the low occur with divergence in at least two out of our three indicators.

We know divergence is a reasonable signal.  We do know that, don’t we?  Divergence on more than one indicator of different lengths is even better.  The Pulsar pattern is a reasonable signal.  The pattern accompanied by divergence in more than one indicator puts the odds tremendously in your favor.  About 60% of the time, regardless of what time frame chart or what market you are trading, you can expect the market to move at least as far in the direction of your trade as your original risk amount (and it will usually move much farther than that).