The main aspect of the work that is presented here is the choice of the rule under which timely changes in the allocations of account equity are decided upon.
The most important understanding that we can have, if we are to fully benefit from quantitative analysis of the trailing history of the markets, is that having a “theory” as to why a particular rule is better than another is hardly necessary. It’s necessary to confirm that an inefficiency exists; the cause need not be known. Empiricism is the appropriate ultimate resort and we should not consider a priori assumptions about human behavior in the marketplace to be trustworthy.
For example, some academicians admit that momentum works, that to some extent recent advances in stock prices tend to be followed by continued gains. That’s if “recent” refers to the trends of the last few months. But if we pay attention only to the trends of the last few days then momentum decidedly doesn’t work; if anything, betting against such short-term trends would work. And none of that fits into the theoretical concept of an efficient marketplace which is taken to be the inevitable result of every investor rationally going about the business of optimizing risk-adjusted returns. While it may be interesting to speculate about how the market’s internal dynamics come about, developing a successful allocation scheme does not require any understanding of the why behind whatever happens.
Presently all of the rules for making changes in allocations of account equity that are in use within the Traded Portfolio project incorporate what is usually called a “walk-forward” procedure— one or more parameters that affect the allocations are determined on the fly, based on the trailing history. In that way we stand a chance of the program proving to be adaptive to changes in the way that the market trades and we may be afforded a major advantage over the more static approaches. And in the process we are automatically testing out-of-sample and thereby arriving at unbiased estimates of performance.
Presently the rules whose outcomes are presented in these pages are based on dividend-adjusted price histories. However, nothing about quantitative financial analysis limits us to using price data alone. And with the present program factors not directly involving price such as value, market capitalization or industry-group classification can be taken into consideration as selection criteria for candidate securities lists out of which portfolios are formed.