Timing Tops: Lessons From Bubbles Past

THE UNBOUNDED INDICATOR
One complaint often heard about ROC is that it is unbounded, meaning it has no upper or lower limit. Several indicators, such as stochastics or relative strength index (RSI), introduce steps into the calculation process to eliminate this problem. But the fact that there is no limit on ROC really isn't a problem. It is an oscillator, just like stochastics or RSI, and it will show when markets are oversold or overbought, just like those indicators will. By imposing an upper limit of 100 on the indicator and a lower limit of zero, users of RSI can arbitrarily assume that a reading of 70 means a market is overbought or a reading below 30 is oversold. Test results show these levels work sometimes, but they are also just as likely to show that a trend is strong rather than indicate the trend is due for an imminent reversal.

Without an upper or lower limit, it is difficult to tell when the ROC has moved to a level that is too high or too low. This issue can be addressed by adding Bollinger bands to the indicator. Bollinger bands are an extremely useful trading indicator and are commonly placed on price. They also work very well on indicators. As a dynamic tool, they are not constrained by the bounds artificially placed on indicators like the RSI.

Using standard deviations allow the price action of the market to dictate what is too high or too low, given its recent history. Adding Bollinger bands to an indicator does the same. If the indicator moves above the upper Bollinger band, it is too high, meaning it is more than two standard deviations above the recent average level for the indicator. A move below the lower Bollinger band shows that the indicator has fallen to a level that is two standard deviations below the recent average.

OR IS IT JUST RIGHT?
Standard deviations are derived from a normal distribution, and assume a random relationship between consecutive data points. This is true when you roll dice, since the next roll is not dependent on what the previous one was. For this kind of data, two standard deviations will contain about 95% of the available data. Since only 5% of the data would fall more than two standard deviations from the average, those are rare data points and we'd want to pay attention since they would indicate that something important is likely to happen. Normally, the data almost immediately starts moving back to the average, which is known as reversion to the mean.

Stock prices don't follow a normal distribution, but the assumption offers a useful stating point for analysis. Bollinger bands are usually placed two standard deviations above and below the average. When prices move above the upper band, it indicates that something unusual is happening. In stocks, we often see a reversion to the mean as prices start falling back to the moving average. In futures markets, except for stock indexes, a break of the upper band is often associated with a strengthening uptrend, and rather than betting on falling prices, a long trade could be initiated.

Stocks and futures generally behave differently for a variety of reasons, and this is just one example of how you need to know what you're trading before putting real money at risk.

The daily chart of silver is shown in Figure 1. The ROC broke above the upper band on April 19, 2011, with the price near $44, just as silver was entering its runup toward $50. More important, the ROC broke back through the upper band to the downside on April 26 near $45, but only days before the top. This may not be a long-term top, but since prices fell by more than 25% from there, it is at least a significant top.

FIGURE 1: TAKING SHORT-TERM PROFITS. On this daily chart of silver, the ROC broke above the upper band on April 19, 2011, with the price near $44, just as silver was entering its runup toward $50. More important, the ROC broke back through the upper band to the downside on April 26 near $45, but only days before the top. Since prices fell by more than 25% from there, this is a significant top.

The weekly chart (Figure 2) shows a similar pattern for the ROC indicator. Long-term traders can even look to monthly charts, as shown in Figure 3.

FIGURE 2: SELL SIGNALS ON WEEKLY CHARTS. Here you see a similar action in the ROC with respect to Bollinger bands as you did in Figure 1.

FIGURE 3: APPLYING IT TO MONTHLY CHARTS. Applying ROC and Bollinger bands on a monthly chart can help to identify significant turning points.

This indicator behavior is not limited to silver. We saw the same thing in cotton. The monthly chart in Figure 4 helps frame the extent of the bullish move.

FIGURE 4: IT'S NOT JUST LIMITED TO SILVER. Cotton surged higher in the past few years with few traders noticing the profit opportunity.

FIGURE 5: COMBINING MULTIPLE TIME FRAME ANALYSIS. The top in cotton prices is clear in hindsight, but ROC helped spot it before price action confirmed the end of the bullish surge.

Bollinger bands are overlooked as a tool on indicators, and the ROC is an overlooked indicator. Combining the two can lead to profits in futures or stocks. We'll wrap up with two more charts and an example of this idea applied to stocks. The QQQ, the exchange traded fund that tracks the NASDAQ 100, shows that the ROC has reached an overbought extreme on the monthly chart, which can be seen in Figure 6.

FIGURE 6: COULD THIS BE A SIGNIFICANT TOP? The ROC just broke back below its upper Bollinger band after an extended run upward.

The daily chart of the QQQ, which can be seen in Figure 7, confirms the bearishness of the ROC. Seeing the same signal occur in multiple time frames adds credibility to the signal. Like many futures markets, stocks are signaling that caution is warranted at this time.

FIGURE 7: THE BEARISHNESS IS CONFIRMED. Seeing the same signal occur in multiple time frames adds credibility to the signal. The daily chart of QQQ includes a bearish signal from the ROC.

RELATED READING
Bollinger, John [2001]. Bollinger On Bollinger Bands, McGraw-Hill.