Recent content by PredictorX

Your input space contains 2^300 possible distinct values, yet you provide 3000 (< 2^12) exemplars? Any number of functions will generate your mapping. This is a fool's errand.

I use MATLAB- not for trading, but for data mining at a global bank. It is my tool of choice. I've used SAS quite a bit, but avoid it if at all possible.

Thanks. Data Mining in MATLAB is mine, and I co-author Data Mining and Predictive Analytics with Dean Abbott. -Will

I have already responded to this post, but remembered afterward that I had written a quick guide to utilizing neural networks here: Family Recipe For Neural Networks -Will Dwinnell Data Mining in MATLAB

That was my point: for model development data, there are no hard and fast rules. I have no idea what type of data or software tools you have available. Asking for specifics like this is like asking "How many bricks will I need to build a house?" I don't know, how many rooms do you want? Do...

The 9 samples in a single dimension is just used to illustrate the point. The minimum required samples for constructing a predictive model depends on the modeling algorithm being used, the nature of the data itself, the attribute selection process, what performance is the minimum accepted...

This will depend on any number of factors, but a significance test of confidence interval can be used to assess the quality of performance measurements on the validation/test data. The Usenet comp.ai.neural-nets FAQ includes some guidance on this, but I'd recommend picking up a book, such as...

As was suggested, a logistic regression could estimate such a probability, but really any classifier (discriminant analysis, neural network, tree induction, etc.) would do. The real trick, obviously, would be in getting the discriminatory power to a level which would make the thing profitable.

Whether this is a route to a successful investment strategy, I cannot say. However, estimation of the probability of the bankruptcy of firms (and individuals) is done using a variety of techniques (logistic regression, neural networks, etc.).

The correct way to normalize data for this purpose is to do it the same way that you did during training. For example, if the data is normalized by subtracting the mean and then dividing by the standard deviation, then the mean and standard deviation are calculated on the training data, and...

While neural networks are certainly useful tools, there are other nonlinear modeling algorithms which are much easier to program (like k-nearest neighbors) which can also be useful. I am not recommending one algorithm over another, just suggesting that some algorithms might be easier to start...

Consistently? Over an extended period of time? After transaction costs?

Can you explain what you mean by "a mass-univariate approach"? Thanks, Will

...or you could actually address my question?

Let's see how well this logic works in another arena: A machine can't be stronger than its mechanical design that was constructed by a human. Hence it can't be stronger than a human. No, that doesn't make much sense, does it? The simple fact is that computer systems have been built which are...