The major question I was looking at in my Market forecast this past weekend was whether stocks had hit the top end of a broad trading range and the Market answered with a sharp sell off today. One hint of an approach to at least a short-term top was given by the Relative Strength Index (RSI).

I don't want to overplay how much the RSI was flashing an overbought 'signal' as this indicator had only gotten close to readings that suggested an overbought market. However, the S&P 500 ALSO had traced out a possible double top. More on just my main (RSI) topic shortly.

First, a look at a key index chart so I can comment on the current market in a general way technically. The S&P 500 (SPX) and the Dow 30 (INDU) were leading the market higher on the recent rally, and the broader SPX becomes the bellwether index to look at for the earliest warnings of a downside reversal.

Given that SPX had run up 118 points from trough to peak from 7/1 to its highest intraday and closing high at the beginning of this week (8/9), couldn't pierce its prior high over several sessions and was quite close to registering an 'overbought' level on the Relative Strength Index (RSI) for several days running, the chart/indicator picture here offered enough reasons to take call profits and run. During the aforementioned several days of churning just under a prior top, there were also relatively high bullish sentiment levels.

I almost never (anymore!) try to hang in for what may prove to be the last bit of a move, up or down; 100-110 points in SPX is a darn good profit provided I bought its options 'right'; i.e., early in an emerging trend.


I'm writing my recent 'basic' series on technical analysis principles, chart patterns and indicators to bring these basic articles into our current (Trader's Corner) database so they are accessible to you and so that I can reference such prior articles when writing about a topic that can be further studied online. The charts seen from this point on are past stock and index chart examples, not current ones; some, from my (Essential Technical Analysis) book.

The Relative Strength Index, usually referred to as the 'RSI' (as is also true of stochastics) is an oscillator which means that the indicator is constructed in a way that its numerical scale goes from a fixed point on the low end (0) to a fixed point (100) on the upper end. Oscillators always have a scale from 0 to 100. This is not the case in the Moving Average Convergence Divergence or MACD ("mack-dee"), which will be the subject of a further article in this series.

The RSI was developed by a trader and market analyst named Welles Wilder back in the 1970's. A simple way to understand the RSI is that it is a ratio (one number divided by another) that compares an average of up closes to down closes for X number of days; there is only this ONE variable that we can normally set in our charting software, which is the length setting or the number of periods (hours, days, weeks, etc.) that the RSI formula works with. A common RSI default (the preset value) for RSI 'length' is either 9 or 14. The relative strength index calculations will be based only on the number of closes specified as the length setting. The reason for the widespread use of either 9 or 14 is mostly a matter of convention. Obviously, a setting of 9 or 14-days does not represent an even number of 5-day trading weeks.

However, both 9 and 14 are common default settings and there is a repository of experience among users of the RSI with these levels in remembered instances of overbought or oversold conditions associated with them. The commonly used default settings are 30 to suggest that a stock or index is oversold and 70 and above to suggest an overbought situation.

The optimum length settings are a function of your time horizons relative to trading or investing. 5 to 9 (or 10) for the RSI length on a daily or hourly chart is appropriate for the minor trend seen in short-term trading. A length setting of 14 up to 21 for daily charts is good for looking at the intermediate-term trend, such as over 2-3 weeks or more. On weekly charts, my preference is to set RSI 'length' to 8, which of course measures an 8-week period; 8 weekly bars represents a 2-month period or 1/6th of a year, which can provide a relevant picture of the secondary trend.

RSI is derived by calculating the average number of points gained on up days, during the period selected (e.g., 14), then dividing this result by the average point decline for the same number of bars – this ratio is "RS" in a 14-period formula for RSI or 100 – 100/1+RS. RS equals the average of 14-days' up closes divided by an average of 14-days' down closes. 13 or 21, part of the Fibonacci number series might be used instead of 14 in this example. I make almost exclusive use of 5, 8, 13 or 21, all fibonacci numbers, as my RSI length settings.

Every up close during this period is added and this sum is divided by the number of 'bars' that had up closes for the number of bars or trading periods being looked at; this, in order to arrive at a simple average. Every down close during the period selected is added, then this sum is divided by the number of bars that had down closes. If 10 of 14 days had up closes, the result of this division is a ratio that rises rapidly. Subtraction from 100 of the result of the division is what makes for a normalized scale of 1 to 100.

In a period of a rapid and steady advance the RSI will reach levels over 70 rather quickly, which is defined as 'overbought' and the RSI can then remain above this level for extended periods if prices then churn sideways. In a period of a rapid and steady decline, RSI can dip under 30, defined as 'oversold', for extended periods.

Once the RSI retreats from high levels, it can offer an alert that the trend may be reversing. It is also true that relatively brief sideways consolidations or even a minor counter-trend movements will cause the RSI reading to back off from the preset 70/30 'extremes' and get back below 70.

The reverse situation would be where a steep decline causes the RSI to fall well under 30; a brief sideways move or counter-trend rally can put the RSI back above 30 fairly quickly. Once a strong trend continues, the RSI can quickly get back into oversold or overbought territory again.

In a strong up trend, it is often advisable to both use longer length settings of the RSI, such as 13-21 on daily charts and to define overbought as a level ABOVE 70; e.g., in the 80-85 zone. In a strong down trend, changing to a longer RSI length setting may be appropriate, in addition to 'defining' oversold only as an RSI level that is at or below 20 for example.

You can change your conception of constitutes the overbought and oversold extremes to an area closer to what has been reached already as the highest, and lowest, RSI readings:

The chart above is also an excellent example of a bearish price/RSI divergence that set up on the second rally to the highest peak. You'll notice that the RSI did not then ALSO move to a similar new high, resulting in a divergent 'sell signal'.

A common situation is where there is a strong initial price swing, then a period of consolidation such that the RSI stays within the 30 to 70 levels. While these readings are in this zone, the RSI is not of great practical use by itself. If the RSI levels hover just under 70 such as seen recently in SPX (first chart above) AND there is a potential double top or some OTHER bearish chart pattern, the RSI is telling for a possible top, but isn't definite according to our usual overbought level(s). Good value is also gained in using the RSI in a market that develops a well-defined trading range such as seen in my second chart above (IBM).

If prices falter in an area of prior highs, accompanied by even 1-day when the 14-day RSI is over 70, this indicator gives an added indication to sell in a prior resistance area. In the area of prior relative lows, accompanied by even 1-day with an oversold reading of the RSI at or under 30, the RSI provides some technical reinforcement for buying while prices are in a likely support zone.

The use of indicators to confirm other technical developments relating to prior highs and lows or such things as price reversals at trendlines, is one of the RSI's most useful aspects. A rounding or rectangle bottom, coupled with an oversold RSI reading, would be good technical validation for bullish option strategies, especially if prices also broke out above a trendline.

Trendlines may also be applied to a series of lower RSI highs or higher RSI lows, with the same implications for a breakout above or below the line; this may lead or 'confirm' similar price action.


Indicators like the RSI, besides their usefulness in providing some idea of whether a market is overbought or oversold, will also generate occasional divergent buy or sell 'signals', on either an intraday, daily or weekly chart basis.

RSI will normally move to a new low or new high when prices do the same and this is bullish or bearish confirmation by that indicator. When confirmation DOES NOT occur, the resulting bullish or bearish divergence is an important alert for the possibility that the indicator is signaling an upcoming reversal to the current trend.

In a sense, such divergences are more 'important' than indicator confirmations as they can suggest either an excellent buy or sell signal. If the signal is one that indicates a possible reversal to a trend that you are participating in, this is a valuable advance warning. One of the chef risks to potential profit is being caught in a trend reversal that you did not see coming and were not prepared for.

A classic example of a bearish divergence such as seen above, is a higher price peak, accompanied by a lower low in the RSI reading, compared to its last extreme. Such a development suggests some likelihood of a downside reversal developing in the near future.

A classic example of a bullish divergence would be a higher high in the RSI relative to what came before; i.e., in the same period when prices made a lower low. Such a situation suggests the possibility of an upside reversal.

There is a third situation that comes up, which is not considered a 'classic' price/RSI divergence, where the indicator bottoms or tops in the same area, marking a double top or bottom or just repeated highs or lows in the same area (i.e., a line formation), whereas prices go on to make a new high or low. This type of divergence is also valid in terms of suggesting that the market may have topped or bottomed. Such measurements of momentum based on an oscillator formula, suggest that the trend strength could be waning and that the item or financial instrument in question, vulnerable to a reversal.

The RSI has a maximum reading of 100 and it has a tendency to 'flatten' out in very strong trends. This is even more pronounced in the Stochastic indicator. However, the RSI will fall relatively quickly from its most extreme point as soon as a correction develops. When a rally resumes, the RSI will rise again with the advance, but will also give a clearer cut divergent reading by not rising as far as previously if the momentum is not as strong on the next advance.

There are often multiple divergences and some will be early and well before the actual occurrence of a trend shift. Instances of multiple divergences create a risk of premature trade entry one or more times if trade entry was based primarily on a divergence between price action and an indicator like RSI. There are some ways to avoid this situation, as follows:

1. Give more weight to divergences that occur during or after overbought or oversold oscillator extremes that have been adjusted to reflect the strength of the dominant trend; e.g., if in a very strong up or down trend, the appropriate RSI overbought-oversold levels may be 80 and above or 20 and below, rather than 70/30.

2. Look for 'confirming' price (chart) patterns that are associated with trend reversals, especially when an up trendline is pierced on a decline or when prices rally and break out above a down trendline.

RSI tends to be my oscillator of choice in looking for divergences, although it can be advantageous to also look at the other common oscillators like Stochastics and MACD. Sometimes it is a different indicator, such as the stochastics model, that highlights a particular bullish or bearish divergence.