"The Band, Elwood"

CXO Advisory takes an interesting look at Bollinger Bands here. They go from 1950 to June study a strategy of using differing BB Band violations as a trading strategy, namely they use a 21 day MA and 1.5, 2.0 and 2.5 Standard Deviation violations. And then hold for 21 days.

In other words, let's say it breaks a downside band. They buy and hold for 21 days.

Just one problem; violations come in bunches. So CXO limits it to one signal per 21 day period, so you can't use a Dysktra-ish strategy of going long 800% of your capital.

And the results?

The 2.5 STD did not prove to be a good signal, so they went further on the 1.5 and 2.0. I'll let them do the 'splainin.

The trading strategy consists of buying at the close after each BUY signal and selling at the close 21 trading days later, thus generating 301 (234) round-trip trades for the 1.5 (2) standard deviation bands. Ignoring trading costs, return on cash when out of the market and tax effects, we find that:

The strategy as applied to the 1.5 standard deviation Bollinger Bands generates a 1105% return over 57.5 years. This strategy is out of the market about 56% of the time. It may have produced substantial net interest income while out of the market, but also mostly short-term capital gains (it involves an average five round-trip trades per year).

The strategy as applied to the 2 standard deviation Bollinger Bands generates a 757% return over 57.5 years. This strategy is out of the market about 66% of the time. It also may have produced substantial interest income while out of the market, but again mostly short-term capital gains (an average four round-trip trades per year).

Buy-and-hold generates a return of 9053% over 57.5 years, winning easily because the stock market still goes up on average during the months the Bollinger Band strategy is out of the market (and compounding of such returns is powerful).

In summary, Bollinger Bands for the aggregate stock market do provide some sense of overbought/oversold (and future volatilty), but they do not as standalone indicators support a trading strategy that beats buy-and-hold.

Anecdotally, I would have to agree; the *extremes* don't really stop anything. What I find with BB Bands maybe is that when it is riding along one band for some time and then escapes and goes towards the MA line, that MA line becomes excellent support/resistence. But that's sort of neither here nor there, I have no objective definition for "some time".