There are many systematic trading plans that can produce results that are better than Buy and Hold. In the search for the top trading plans, a statistically sound approach should be used. Past performance is certainly no guarantee of future results. There are extreme factors that occur that may affect short term analysis. It is possible that there may be changes in markets characteristics that can also make historical analysis less valid to current market conditions. There is a saying that the more things change, the more they remain the same. Market timing requires discipline to take every trade.
Approaching systematic trading research in an objective and scientific manner can provide the confidence to stay with a trading plan. But staying with the trading plan means to me, that you stay with it until the evidence tells you otherwise. Companies like Walmart, Microsoft and other large company’s trade differently after early stage results than once they become mature companies with growth rates that are now much different than 10 or 20 years ago. I would use a different market timing system today for Microsoft than I would have used in the 80’s.
Back-testing over several years and in different markets is critical for a good market timing system. Just using Bull Markets or Bear Markets will invalidate scientific research. The 37% decline in the S&P 500 in 2008, I hope is not the normal year. The frenzied capital expenditures of 1999 into year 2000 may not be typical either. Both periods, as well as Black Swan events like September 11, 2001 should be included. Technical analysis methods on hindsight may not have told you what was coming, but often can be looked back on as an excellent predictor that something was changing.
How is this research in a scientific manner accomplished? It needs to use a large sampling of data for a variety of assets. The data to obtain evidence of the best market timing system needs to be fed into a computer and the results analyzed. Remember, we are always talking about back-tested data and therefore historical results that may or may not repeat themselves with consistent reliability. The systems tested must have specific programmable entry and exit rules. Only rule based trading that can be written for a computer that has no room for interpretation should be tested. The input parameters should be far enough away to give different results. For example, the testing of a 50 day moving average against a 55 day moving average is not a significant different test. The testing of a 50 day simple, an 80 day simple moving average, a 100 day and a 200 day moving average would be more appropriate.
Computerized algorithms systems allow this process to be completed and the results ranked by profits. A combination of trend following approaches, swing trading approaches and other statistically sound approaches may be put into the computerized algorithms. Each day, a trading signal is generated by the computer. In testing a trading system that you are always either long or short each day, the only results the computer will produce is “Stay Long”, “New Long”, “Stay Short” or “New Short”. Some assets trade better using Trend Following systematic trading plans. Other assets and indexes seem to have reversion to the mean characteristic and should therefore be traded accordingly.
With the evidence of the past, this information can provide some insightful findings that may be used to trade these models in the future. Evidence based research is used in medicine, science, and other areas where specific procedures and operating procedures need to be established. Scientific methods with statistical inference may be the best tool for portfolio management of a large index based portfolio.
For more on systematic trading visit www.seleznovcapitaladvisors.com
