EdgeWare FastGraph Version 3 - Manual (Page 21)

21
Finally, what are the best parameters to use for buy and sell ranking? The values that
gave the best annual return may not be the best choice because that combination may
have undesirable characteristics, i.e., high drawn down, too many switches per year, etc.
The most significant problem - the parameters may not be a good “robust solution” and
only a “statistical fluke.” Stability is a requirement in your parameter choice. In other
words, if the parameters are changed slightly, there shouldn’t be a large change in annual
trading performance. You do not want to “curve fit” the past. In the above example
holding three funds may have resulted in a slightly lower return, but the return was quite
consistent and peaked in the 25-30 day ranking period. Holding a single fund shows con-
siderable variation in performance, For example, ranking over 24 days has over a 35%
annual return but ranking over 16 days results in less than 20% annual return.
Using FastBreak with different ranking periods for buying and selling the problem be-
comes a little more difficult because there are now two independent parameters and many
more combinations. Contour plots can be used to help select “optimum” parameters. For
example, make a FastBreak run the same as Run #1 earlier with the following exceptions:
Buy min and Sell min set to 10, and Buy max and Sell max set to 50.
Import the file in FastGraph, select contour as the graph type, and Avg Annual % as the
Date Field. You get the following graph:
Note: The graph will be in color on the computer monitor.
The annual return is shown as a different color and is a function of buy and sell ranking
periods. It appears there are large regions of stable/robust parameters. Parameters are
considered robust when by changing them by a moderate amount the results stay consis-
tent.