Endnotes

Chapter 4

1. See Garfield Drew, New Methods for Profit in the Stock Market, 1955, reprinted by Fraser Books, Burlington, Vt, for a discussion of some famous investment plans.

2. Certain highly adaptive systems may be exempt from this dictum, at least to some extent. Fuzzy logic, neural networks, and genetic algorithms are tools that may well allow a system to be adaptive enough to survive. See www.EquityTrader.com for an example.

Chapter 5

1. Eventually broad concepts may achieve such widespread acceptance that they become diluted or perhaps even inverted in extreme cases. Witness the expected long-term outperfor-mance of small-capitalization stocks, a concept that became so widely embraced it vanished completely. In the parlance it is said that it was "arbitraged out of existence." However, this may only be true for concepts that achieve true mass acceptance-down to the "man on the street," as they say.

Chapter 6

1. The history of technical analysis is obscure at best. I have endeavored to be as thorough as possible. Much of the original material was published in newsletters or ancillary documents. Many source documents are simply lost or may have been deliberately destroyed. Much knowledge and many techniques were kept secret, and many of the cognoscenti have passed on, taking their "secrets" with them. Claims are often conflicting, and those who might be able to resolve them frequently have an ax to grind. Any additional citations on bands, envelopes, and related indicators or systems would be greatly appreciated.

2. The typical price is a very old technique. The most commonly used formula is (high + low+ close)/3. We would extend it to (open + high + low + close) /4 when the opening price is

Endnotes

available. The use of the typical price to compute the Bollinger Bands is a technique I recommended early on, one that is still quite useful. Using the typical price as a base for Bollinger Bands results in slightly slower, somewhat smoother bands, which may be an advantage in certain applications.

3. Richard Dennis, a famous commodity trader, taught a number of traders his proprietary techniques. Those traders are called Turtles.

4. Hurst, now a recluse, gave a series of seminars in the 1970s in which he expanded greatly upon his work. Much of this material was thought lost until recently when Ed Dobson of Traders Press in Greenville, South Carolina, contacted Hurst and undertook a project to republish the CycliTec Services Training Course. Originally published in 1973, the course consists of three massive binders and a set of audiotapes.

5. Stock market breadth refers to market statistics such as advancing and declining issues, the volume of issues up or down on the day, and new 52-week highs and lows. These measures are said to depict how broad the rally is-the theory being, the greater the participation the better.

6. It is this squaring of the deviations from the average that makes Bollinger Bands so adaptive, especially to sudden changes in the price structure.

Chapter 7

1. It turns out that historical volatility and projected or implied volatility are related, and the differences between them contain useful trading information.

2. The population calculation is used for Bollinger Bands, not the estimate calculation for which the divisor changes to n - 1. There is no theoretical reason for this. In initial testing the population seemed to work well, and so it was used. The bands would be a bit wider using the estimate calculation.

3. There is another reason, one that is beyond the scope of this book, but which Ill touch upon briefly. Securities prices are not "normally distributed"; they are more variable than one would expect. This is known as having "fat tails." Thus many statistical inferences dont hold. For example, at 30 periods

Endnotes

the bands hold near 89 percent of the data for stocks, rather than the 95.4 percent one would expect. See Chapter 9. 4. An 85 percent containment was initially suggested by Marc Chaikin for Bomar Bands and has proved to be a very useful quantity. So we were pleased to find the containment in these tests running in the same range.

Chapter 8

1. Those of you familiar with technical analysis may recognize George Lanes formula for stochastics as the basis of the above formula. The basic stochastics formula is (last - n-period lowest low)/(n-period highest high - n-period lowest low). Mr. Lane named his indicators %d, %k, etc. So in accordance with his system, and in acknowledgment of the derivation of my approach, I followed his naming scheme and used %b after checking to be sure that Mr. Lane hadnt already used it. George Lane comes from the Midwest, where he barnstormed for many years in a Cessna 210 teaching his trading approach to commodity traders, primarily farmers. He comes from a deeply religious background, and his lectures have the flavor of the revivalist tent. He is not to be missed! He was the first to teach me the idea of three pushes to a high-in a style that I will never forget. It is a concept I have come to appreciate evermore as time has passed. He uses stochastics to diagnose the three-pushes-to-a-high concept; I use Bollinger Bands and volume indicators. I often wonder what indicators he might have come up with had he been more focused on stocks, where volume is concurrently available with price, instead of commodities, where volume is reported a day late and as a single, estimated point.

2. The mathematically inclined will note that using the default settings for Bollinger Bands BandWidth is equal to four times the standard deviation divided by the mean, or four times the coefficient of variation. Again I make no statistical claims on this basis. I only point out the derivations of the calculations we use.

3. Mr. Cahen is also an advocate of the use of three time frames. His approach, which he refers to as his "triptych," consists of