Tactical stock trading
by Peter Eliason
We have all heard that there is no solution to predicting the market because price movement is a
random walk. The statement is partially right and partially wrong. Price movement may be random, but
there is an exact solution to a random walk that can be used to mathematically beat the markets.
The solution revolves around the use of a tempered martingale numerical series. This algorithm is not
probabilistic, but it is mathematically exact. (See Stocks & Commodities, July 1988, page 40.) It has two
components: a point spread that determines exactly when to buy and sell and a numerical series that
determines how many shares to buy or sell to produce a pre-known gross profit.
We start with a beginning numerical series such as [1 2 3 4 5 6], with each number representing a
"factorable" number of shares to own. The "factor" is multiplied times the number in the series to
determine the number of shares which will be traded. For example, a factor of 100 would make the "5" in
the series represent 500 shares.
Once we have the series, we need a "base price" which is usually just the price of the last buy, sell or no
action transaction. We start by buying at the base price and set two limit orders a point spread above (a
sell) and below (a buy) the base price. If a stock's price is $10 per share, the limit orders may be set at $9
or $11—or any other spread, for that matter.