Support MarketTrak, Visit Our Advertisers


 MarketTrak Question/Comment Message


Posted By: Tom D   Date: Mon Nov 17, 2008

Title: Change

Message:
  Thanks for your honest answer. I understand the difference between back propagating and evolutionary networks. I also see that your networks haven't held up very well and have necessitated frequent changes.
 
  My own experience 15 years ago led me finally to discontinue all the work despite initial great success. Data inputs are far and away the most important factors. It may be that data series searches need to be as computer intensive as network training and selection using old data. Markets do not always simply respond to the same data inputs over time. Traders and fundamental analysts know this or they die. Data series that used to work are sometimes no longer relevant. What are the new data?

Response:
  Tom, I agree with your comments on using standard networks. The networks that I use are far from standard. These nets have logical branching and adaptive functional creation. There are 63 inputs that are computed from the data types that I showed in your previous post. These inputs are moving averages and basic technical and fundamental indicators. The networks take these 63 inputs and internally generate functional forms that best model the market over about 5000 days of data. This data is the training set. The most recent two hundred days of data are not included in the analysis and therefore provide a blind test set. The functionals that are created are very complex and I don't fully understand them. Each network in the library of some 200 is different which increases the complexity of the overall system.

  If you need more in-depth technical info, it might be best to email me.

Return to message board




 

Copyright © 1996-2024 MarketTrak LLC.  All Rights Reserved.