|
MarketTrak Question/Comment Message |
Posted By: john Date: Tue Jun 30, 2009 |
Title: the models could all be wrong? |
Message: |
Response: Let me give an example. Suppose the accuracy of each model is
70 percent. Suppose further that they are completely independent of
each other and they look at datasets which are not correlated. Then
the chance that all three could wrong is less than 3 percent. Of course,
the models are derived from the same basic model and they see the same
market data so the probability of error is much higher. |