This page intentionally left blank.

R

NORMALIZING INDICATORS

Bollinger Bands need not be confined to use on security or index prices. To name just a few of the possibilities, they can be used on ratios, economic series, fundamental data, volume, and technical indicators. It doesnt matter whether these series are presented as oscillators or as series without bounds. In each case Bollinger Bands serve the same function they serve with price; they define high or low on a relative basis. This can often add great insight that the traditional rigid levels and rules cant deliver.

Take the Relative Strength Index, RSI, as an example. The traditional rules call for an overbought interpretation when the indicator rises above 70 and an oversold interpretation when the indicator falls below 30. Perhaps the most commonly employed technique using this indicator is to buy a positive crossing of 30 and sell a negative crossing of 70. This, however,

Part V: Advanced Topics

can be problematic. Sometimes 70 and 30 work, sometimes they dont, and sometimes they fail spectacularly.

For a more robust approach to RSI, take the observation that in a prolonged bull run the 70-30 decision frame rises, with some analysts recommending 80 and 40 as the benchmarks, and that in a prolonged downturn the frame falls, with 60 and 20 being recommended benchmarks. Indeed, a tag of 80 can be used to define an uptrend, with 40 being used thereafter as oversold until a tag of 20 reverses the structure and suggests a new downtrend, with 60 now being overbought. Thus the action of the RSI indicator can be used both to define the major trend of the market and to shift the decision frames to the appropriate levels to identify overbought-oversold relative to trend.1

A bit of a digression is in order. In the old days, overbought and oversold were terms used to denote climactic conditions. In the fall of 1974, at the end of the last great bear market, the stock market was oversold; and in the spring of 1962, as the bowling stock craze2 crested, the stock market was overbought. Analysts looked for these phenomena to occur infrequently and to mark important long-term turning points in the markets. The passing years have seen the relentless contraction of the time frame; these definitions have become moot for the typical investor. Today overbought and oversold are applied to the very shortest of time frames, a practice that no doubt earlier analysts would find simply incredible. However, we must accept the basic definition as applicable to all time frames: to have come too far, too fast.

While the shifting of the decision frame for RSI according to market conditions is a clear improvement over the simple 30-70 interpretation, we can do better still. How? By plotting Bollinger Bands on the indicator and using the bands to set the overbought and oversold levels. By using Bollinger Bands in this manner, we obtain a fully adaptive approach that morphs with the market. First a bit on setting up Bollinger Bands on indicators; then a neat trick.

Indicators are a more diverse lot than stocks. Unlike stocks where 20 periods and 2 standard deviations are the best departure points for the vast majority of issues, each indicator seems to require its own parameters (see Table 21.1). In general, the appropriate averages tend to run longer than those for stocks, a fact already acknowledged in our use of a 50-day average for