Mechanical Investing is using computer programs to test, analyze, and develop working models for stock selection. One selects various financial and relative strength criteria to develop a unique set of instructions that can then be entered into a computer program capable of going back at least a decade while applying the instructions to the data that existed back then for selecting stocks. In it's most simple form for example, one may think that only stocks that have a ROE(return on investment) that is increasing x% over time and is greater than 15% will go up much faster than other stocks. Also, one may believe that the relative strength(price performance) of those stocks with high and increasing ROE's must be at least 80%. Mechanical Investing then would write a series of steps to take and then apply those steps to prior years to test whether stocks selected based on these specific criteria existing at that time in the past, would have produced superior returns.
It is important to have a very large data base of stock prices and stock fundamentals and basic financial data from the financial statements of all companies. The backtester scans the universe of stocks that had ROE's over 15% and have grown x% or more each year covered by the backtest, and from that group only selects those stocks which have a relative strength of, say, over 80, if that is what the original instructions give us.
One can buy and hold annually, quarterly, semi-annually, weekly, or monthly. After the period ends, stocks not remaining on the list derived from the screener program are considered to be sold and the new stocks that pop up on the list are bought. This is the rebalance date. The process is continued through the entire period and the compound annual growth rate is calculated over the period tested under the backtest.
Based on the results, a statistical determination is also made to determine how volatile the returns are and a Sharpe ratio is calculated for each portfolio. This is done to compare screen results with other portfolios or benchmark markets to examine whether the range of returns and the amount of the returns would probably be higher than other alternatives.
Once results are in, another test can be done using the same criteria but using different amounts (Say RS of 85 and greater rather than 80 or greater) or by adding additional criteria to the mix. One may want to balance certain growth aspects with certain value aspects and develop a "growth at a reasonable price" strategy to be tested. So one may then want to add price to book ratios or other value indicators into the mix.
There are already some very powerful tools available for free on the internet to accomplish the testing. Many folks have given their time and their expertise to developing these tools. If we can talk about our ideas, we can test them and see whether they would have worked and whether they are likely to work in the future, based on a certain statistical confidence level.
We can learn together and discover together what works and what doesn't. We can try new ideas that hopefully make some sense logically and financially.
Some people have already delved into this arena and have written some books about mechanical investing. And these authors took their work only so far. What they learned can be useful and a great starting point. But there is much more to be done and that can be done.
Interested?
HiTech |
