Irwin, thanks for some links. Before I start reading all this I like to mention that I think curve fitting can be used to prove all sorts of falsehoods, like the saying goes: "With Statistics we can prove anything we want to prove"(This does not mean that I lump curve fitting under statistics).
In regards to curve fitting in an investment, I forgot what started this discussion. Is there a reason to assume we disagree on something or is it simply that I asked you how curve fitting is used in investing systems?
I think, for example, that curve fitting on any stock price has no meaning, although it might be true that it is used in investing. If so, then is has meaning in so far the user thinks it helps him. That's fine, but this is, in my opinion, not the purpose of curve fitting.
If you fit a curve to stock prices then most of the real data does not lie on the fitted curve. A fitted curve is intended to represent the real data, and as such a fitted curve will closely fit the real data. The assumption is here that measurements do not represent real data, but due to errors represent the real data approximately. The fitted curve is then assumed to represent the real data more closely than the measurements do(but this is not always true either as spikes in the real data can be missed in the measurements if then measurement spacing is too far apart).
With investing, and data on prices, the reality is different. Stock prices are the real data! Any curve fitting will lead away from the real data.
In the limited sense that a fitted price curve may be useful, for example as a moving average, it does not satisfy, in my opinion, the purpose of curve fitting as it does not tend to characterise the process that produces the real data.
So, in so far as a moving average characterises the average from the past, it does not characterise the real data in the past.
I will peruse the links you provided later to learn more about curve fitting.
Conrad
Market Data 
Markets 
AI
