Racket wrote:No. Why? Because it's complicated and involves a lot of time and math which I don't feel like setting up in R right now...
No Racket, the reason is because it's not possible. You can generate whatever distribution you like to match the data as best you can, but you can not produce better "predictive power" than one can get from a simple examination of the process itself (the past data).
You will not produce a "better" prediction not because it's too hard or you are too lazy, but because it's not possible.
Predicting what will happen in the market today (or any day) is a fool's game, which was the whole point of what I wrote (which basically boils down to the expected change is ~ 0, with a slight edge to a gain instead of a loss).
You, like many / most linear thinkers, seem to imagine that mathematics on their own provide a perfect basis for determinism. Mathematics are beautiful, and they are very often useful as a philosophical substitute for reality, but they are never actually perfectly representative of reality itself. Never. Not even in the case of the most "perfect" physical "laws" (although in some cases they are remarkably accurate and useful).
With financial markets, the weather, climate and many other phenomena, mathematical models are for the most part useless. Prove me wrong...