Linear Programming

[ikicker]

Which metrics should I use in a model when attempting to model stock price based on correlation coefficients with linear programming? I'm attempting to study the correlations of a specific stock to take into account positive news events, negative news events, profits, net income, etc, but I do not know which ones to use in the model. I realize there will be substantial collinearity which will need to be eliminated.


Also it would have to lag a good amount. I guess you could use leading indicators to predict some of a price in the future?.


Thanks.

 

[Rhody Trader]

Read The (Mis)Behavior of Markets by Mandelbrot to see why you're barking up the wrong tree.

 

[r.w.]

Which metrics should I use in a model when attempting to model stock price based on correlation coefficients with linear programming? I'm attempting to study the correlations of a specific stock to take into account positive news events, negative news events, profits, net income, etc, but I do not know which ones to use in the model. I realize there will be substantial collinearity which will need to be eliminated.


Also it would have to lag a good amount. I guess you could use leading indicators to predict some of a price in the future?.


Thanks.

Not sure I understand what you mean by linear programming. To me, that term means optimisation of a linear function given linear constraints. To me it sounds like you're looking for some kind of cointegration model. There's a very readable introduction to such models in "Pairs Trading" by Ganapathy Vidyamurthy, including signal/noise ratios, tests for tradability and so on. Then, again, I may have misunderstood completely. But those models would give you what you asked for - the metrics for trading design.

As to the intended application, I would have to agree with Rhody Trader, again with reservation for being mistaken about what you are trying to say.