|
Parameter Selection |
Top Previous Next |
|
Parameter Selection
The selection of the 'best' parameters to use with the cyclePack indicators at any point in time is, without a doubt, more of an art than a science. If you experiment with the indicators using historical data (use the Anchor button to relocate the point-of-origin) you will quickly find that it is almost always possible to improve the quality of a projection (in hindsight) by either relocating the point-of-origin or by adjusting some of the parameters such as Data-Window size, Resolution, Persistency Filter, etc. When applying the indicators during trading, however, it is unlikely that you will have either the time or the inclination to fiddle endlessly with positioning and/or parameters.
So what can you do to maximize the functionality while minimizing the hassle? Here are some tips and observations:
Default Parameters: The good news is that the default parameter settings were selected as the result of a lot of testing and they work very well across a wide variety of securities and intervals. In most cases you will not need to change them.
Confirmation: The cyclePack indicators provide the most value when used in conjunction with other technical analysis tools that you already use and trust. For example, if both your Elliott Wave count and your cyclePack indicators suggest that a change is imminent, a higher degree of confidence in the projection is achieved. The same logic can also be applied to more traditional indicators such as stochastics, CCI, RSI, etc. Yes, you may run into situations where you miss a good trade (e.g., the cycle projection was spot on but you passed on the trade because your other indicators did not agree) but in the long run it will serve to keep you out of trouble. As you gain confidence in the cyclePack indicators over time, you can begin to place more weight on the projections.
Common Sense: If you find yourself in a situation where moving the cyclePack indicator point-of-origin slightly results in drastically different projections, just pass on the projection. Instead look for situations where the projection is somewhat consistent over a range of bars. The projection from 5 bars ago will never be identical to the projection from the current bar but it should at least be close. Use the Anchor button periodically to slightly adjust the point-of-origin and compare the results to other nearby projections.
Patience: When you find a really good projection, keep it in place until the current price action begins to deviate clearly from the projected path. Don't spend too much time looking for new projections when the one you have is already working well.
Detail: This applies primarily to the cycleProjector indicator. The # of Cycles to Use parameter allows you to adjust the total number of extracted cycles that will be used to create the synthetic time series and, thus, the projection. Using a large number of cycles can indeed result in a very detailed projection (showing every wiggle and turn) but in many cases it can introduce noise and obscure the behavior of more meaningful cycles. In our experience, it is generally better to sacrifice some of that (potential) detail and focus more on the stronger cycles. Keep the # of Cycles to Use setting down to a reasonable number (e.g., between 5 and 10 cycles) and you will be happier with the overall results.
Data: When applying the cyclePack indicators to intraday bar intervals, be sure to at least compare the performance of the indicators using a 24-hour Time Template to the performance when using a RTH Time Template (e.g., 9:30-16:00 EST). In many cases the 24-hour template will deliver more accurate projections.
Copyright © 2010-2013 Divergence Software, Inc. |