"Overbought" and "oversold" are words without precise definition. Nevertheless their connotations are clear enough. Both suggest that price has reached an extreme and imply the promise of an imminent turn. No wonder that various overbought/oversold (OBOS) indicators have become so popular!
Unfortunately, many traders use these indicators in isolation, when proper use requires that they be considered within a larger technical context. In a later lesson we will take up the proper use of OBOS indicators. In this and the next lesson, we will review the logic of several different indicators and build a single indicator which combines the best of each.
The earliest, and arguably still the best, method for gauging whether an issue is OBOS is the simple trendline. When Trend lines are broken, either with the trend (climactic shakeout) or against the trend (counter-trend shakeout), a stock or commodity is OBOS relative to that trend (see Trend lines 1 & Trend lines 2)
There are three other methods commonly used to indicate OBOS. The first of these is the Relative Strength Indicator, or RSI. Over some period, say 20 days (one could as well use any other period--hour, week, month, etc--depending on focus), RSI compares the sum of daily gains with the sum of daily losses.
Suppose that an issue has moved up for eight out of the last twenty days, and that the total gain over those eight days is +4%. Assume further that the issue fell during ten of those twenty days, but that the total of those ten losses is -4%. On two days there was no change.
To calculate RSI, begin by determining the ratio of gains to losses. The resultant ratio must be a positive number. This may be accomplished by changing the sign of the sum of losses, as in this formula for the ratio:
Total Gain/- Total Loss = Ratio
The rest of the RSI formula is given below:
100 - (100/(1+1))
or 50.
RSI is calculated to range from zero (maximum oversold reading) to 100 (maximum overbought). If there are only losses during the look-back period, then RSI will register zero; if there are only gains, the RSI will indicate 100. In the example above, a total gain of +4% is compared to a total loss of -4%. With total gain and loss even, RSI is midway between its two possible extremes, or 50. In that case, the issue is neither overbought nor oversold.
Readings of 30 or below are generally considered oversold, while readings of 70 or more indicate overbought.
The idea behind RSI is simple yet insightful. When enthusiasm for a stock peaks, positive days are likely to overwhelm negative days, in which case the RSI will approach 100. The reverse is true during periods of accelerated selling.
Any period may be chosen over which to calculate the RSI. The longer the period chosen, the longer-term the implications of the RSI reading. For short-term indications, a twenty-day RSI is best in most cases. A forty-to-sixty-day RSI picks up intermediate OBOS conditions.
Stochastic locates the present price within a recent trading range. Suppose that over the last 40 days an issue reached a low of 40 and a high of 60. If the stock now trades at 50, it is midway between the recent high and low of the range. Under this method, a stock trading near the low of its recent range is oversold, but overbought if near the high of the range. Where "Last" is the last price and "Minimum" and "Maximum" are the lowest and highest prices recorded over some period, the formula for Stochastic is:
(Last - Minimum)/(Maximum - Minimum) x 100
Stochastic readings range from zero to 100. OBOS readings are considered the same as for RSI: 30 or less is oversold, while 70 or more is overbought.
The idea behind Stochastic appears reasonable enough, but in practice the method presents the trader with a serious problem. When a stock is rising into new high ground, it becomes, by this method, overbought and subject to sale. But as we have seen (The Nature of Strength-Following Markets) very strong stocks, those making new highs, are often the best buy.
Strong stocks can become chronically "overbought". Traders who rely on Stochastic are likely to avoid the best purchase candidates during strength-following markets and to sell strongly trending stocks far too early. The mirror-image of this problem besets traders who use Stochastic to trigger the purchase of very weak, but "oversold", stocks.
One OBOS method, the trend channel, attempts to remedy the deficiencies of Stochastic by recognizing trends. Unlike Stochastic, where maximum and minimum are determined by the fixed high and low of the recent range, this method creates a channel, or envelope, which tracks the trend. The target is overbought as it approaches the upper limit of the channel and oversold as it dips toward the lower limit.
A moving average is used to smooth price activity and to approximate the trend. In the example below, a twenty-day moving average (MA) is used (shown in red). Two lines are drawn, one above and one below the MA, at equal distance from the MA. These two lines envelope the price action of the target and establish a maximum and minimum for the purpose of calculating OBOS.
|
|
Unlike Stochastic, a trending stock will not become OB or OS as it reaches the limit of its recent trading range. Instead, rising stocks, for example, do not become overbought until the upper limit of the rising channel is approached.
It is the nature of strong stocks not to become overbought easily, but to become oversold on relatively minor corrections or consolidations. The inverse is true for weak stocks. The trend channel method, which takes the trend into account, captures that reality better than Stochastic.
|
Of course, one question remains: how far from the MA should the channel lines be drawn? This is an important consideration since the location of these limits determines OBOS.
One ingenious solution uses volatility to place channel lines. The result is a channel which narrows and widens as volatility increases or decreases.
|
Volatility can be quite variable, and hence the OBOS range can fluctuate wildly. The effect is startling, but the practical utility of this device is questionable. Since the range is determined by averaging recent volatility, and since volatility can change rapidly, the range may narrow just as volatility jumps or widen just as volatility decreases. This produces trend channels which are out of synch with the action of the target.
Trend channels are a promising approach, but the issue of channel width remains. The next OBOS indicator not only solves that problem but also combines the best elements of both Stochastic and the trend-channel method.
The idea is simple. As price fluctuates about a moving average, during some period, say 100 days, price trades at extremes above and below the MA. These extremes vary from period to period and from issue to issue, depending on volatility and the trend.
The Trend-Adjusted Stochastic sets the upper channel line at the maximum price fluctuation above the MA. The lower channel line is set at the minimum price excursion below the MA. In the chart below, maximum and minimum distances from the MA over the most recent 100-day period are shown.
|
The calculation for Trend-Adjusted Stochastic is the same as that for Stochastic given in the last lesson, except that the maximum above the MA is used as the "high" of the recent range, and the minimum below the MA is used as the "low" of the range. Here's how to calculate the Trend-Adjusted Stochastic:.
1. For each of the last 100 days*, subtract the value of the 40-day MA from the price. If price is above the MA, the result will be positive; if below, negative.
Price-MA
2. Divide the result in step 1 by the MA.
(Price-MA)/MA
3. These calculations result in a series of 100 numbers, some positive and some negative. Now employ the formula for Stochastic offered in the previous lesson, using the last number in the series as Last ("last price"), the maximum of the series as the Maximum ("high" of the range) and the minimum of the series as the Minimum ("low" of the range):
(Last-Minimum)/(Maximum-Minimum)x100
*Note: The look-back period used to establish maximum and minimum for the series may be any length equal to or longer than the period used to calculate the MA.
Trend-Adjusted Stochastic (TAG) and RSI are quite different methods for calculating OBOS. Combining the two brings together the merits of both. Pulse is calculated by averaging together RSI and TAG readings.
In practice, if the period used to calculate RSI is half that used for the MA of the TAG, the results will tend to support each other. For example, a 20-day RSI may be combined with a TAG using a 40-day MA to produce a short-term Pulse. An intermediate-term Pulse might use 40- and 80-day periods, and so on.
|