Development of the Meridian Trade Selection Method

Phase five of our research and development effort was to successfully leverage what we had learned in the first four phases and to build the Meridian Trade Selection Method.  This was a long and intensive process.  From our research we knew that for our trading method to be successful it had to be: 

1. Long-term in nature
2. Robust (not curve-fitted or finely tuned to historical market movements – see Appendix F)
3. Average trade profits had to roughly $1,000 or better
4. Well diversified (discussed in Appendix C)

As a result we built a trade selection method that was exceptionally simple and subsequently exceptionally robust.  It analyzes long-term market movements using a single trade decisive variable.  The tables below illustrate the effectiveness of the trade selection method as tested across 33 markets for 25 years.  The 11 parameters along the left side represent different sequential numbers (changed to letters to protect our legal trade secrets) that were plugged-in to the one trade decisive variable.  As can be seen in the “Totals” column, all parameters tested were profitable and parameter “G” was the most effective parameter.  However, more importantly “G” is actually the pinnacle of a bell-shaped curve.  For example, the parameters immediately to either side of “G” (F and H) are only slightly less profitable than “G”, and similarly the parameter to either side of those parameters (E and I) are again only slightly less profitable, and so on to form the bell-shaped curve.

larger view

This bell-shaped curve has significant meaning.  It tells us that if market conditions change such that parameter G is no longer the most efficient or effective parameter to use within the trading method that we can reasonably expect the method to continue to function profitability.

The three charts below illustrate that the bell-shaped curve and the overall performance continues to hold true throughout three distinct testing phases.  The first chart illustrates testing results from our initial development phase in which we used data from 1979 through 1999.  Chart 2 depicts results from a post development test in which we included out-of-sample data to determine if the method would continue to work without the benefit of hindsight (the use of out-of-sample data dictates that the trading method had been finalized using a different data-set and could not be altered to better suit the newer data).  The third chart is an example of tests we ran against each of the Meridian portfolios using purely out-of sample data.  Again, we continue to see the bell-shaped curve, parameter “G” as the most effective parameters, and the overall performance characteristics remain in-tact.