Relative Strength Index - RSI (P)

By: Adam Oliensis

The following is an excerpt from Adam’s new book Stock Chart Analysis & Dynamic Trading System

Despite the name, the RSI does not measure a stock's strength relative to the market or to another security.  It measures the strength of a stock's upward moves relative to the strength of its downward moves over a given period.  RSI is an extremely popular price momentum indicator.  It was derived by J. Welles Wilder and is described in detail in his book, "New Concepts in Technical Trading Systems."

The formula for RSI is:

RSI = 100 – 100/(1 + RS(P))

Where:

RS(P) = the exponential moving average over period "P" of the upward price changes, divided by the absolute value of the exponential moving average of the downward price changes.

That is (in English), RS(P) is the average of upward price changes divided by the average of downward price changes over period "P."

RSI is an oscillator that fluctuates between 0 and 100, indicating an overbought condition when it is above 70 and oversold below 30.  Another method for interpreting the RSI is to look for divergences from the underlying price patterns, and to evaluate the chart patterns RSI forms in its own right.

Because each security has its own volatility properties, it is often effective to vary the period used to calculate the RSI (varying the "P").  Doing so has a dramatic effect on the volatility (noisiness) of the indicator, and can make it more or less useful more or less often.  Of course the tradeoff is that if you make the "P" too small, the RSI line becomes too noisy, and if you make the "P" too large the line becomes unresponsive to meaningful moves.  Some popular time intervals are 9, 14, 21, and 25 day periods, all of which are thought to correspond to different periodicities in stock cycles.

Here's a chart of BRCM using a standard RSI(25)

It's pretty hard to interpret.  RSI never gets up over the 70 line (dotted) or below 30 (also dotted).  We can see that there's something interesting going on, but we can't see it very clearly.  So let's try a different "P" and see what we can see.

This chart of Broadcom (BRCM) gives some good signals using RSI (10).

This chart is a lot clearer and easier to read.

The light grey highlights trace RSI tops in (or near) overbought territory.  The dark grey highlights track RSI lows in (or near) oversold territory.  There's enough on this chart that's meaningful that we could BEGIN do devise a trading system using it.

In addition to the overbought/oversold conditions this RSI line reveals some interesting divergences.  Look at lines A & B, and at lines C & D on the chart.  In each case, Price shows two local highs at roughly equal levels.  However, on the RSI line (in each case) the second spike is lower than the first spike.  These are negative (bearish) divergences.  In both cases, A/B and C/D, we're being told that the strength of the up days relative to the down days was DIMINISHED as the market RETESTED the prior high.  The market had less UMPH as it tried to break above its prior high.  And that lack of UMPH translated into collapse in each case.

Note: If the RSI line showed HIGHER highs on the second price spikes, those would be bullish divergence and would project higher prices.

This is just one sort of divergence that RSI can give us.  The indicator can be useful in a variety of ways.  We can look for breakouts, trendline breaks, for Head & Shoulders Tops and Bottoms, and so on.  These RSI formations can serve as confirmations of—or divergences from—price formations.

However…there are some limitations on RSI's usefulness.

Look at the grey circled area on the chart.  In that area we see one of the shortcomings of relying on just a single oscillator.  During the circled period, the stock was in a STRONG downward trend.  If you'd bought on the RSI buy signal (a cross up out of oversold territory) you would very likely have been taken out of the stock for a loss.

While oscillators like RSI are genuinely valuable to gauge overbought and oversold conditions, and they often give excellent trading signals, there's more to trading well than just overbought/oversold.  Strongly trending markets can become overbought or oversold and REMAIN SO for quite some time.  This is terrifically important to remember as we develop an overall approach to trading.  We have to add tools to our toolbox that will help us effectively gauge the large environment in which a stock moves to local extremes of price.  By understanding that larger context, we can avoid making a lot of bad trades.  (And avoiding bad trades might be even more important than finding good ones.)

It's a good idea to use short-term oscillators in conjunction with tools that can talk to us about the underlying trends.  That way we can avoid situations, like the one circled above, in which the blind use of RSI(10) would have, during one period, gotten us long BRCM repeatedly and against a very strong trend.