J.M.HURST

[JLS]

TREND1,

you wrote "Future is unknown. IMHO".

So give it up if that is what you truly believe.

It's not that simple. The future is never 100% known, but it is somewhat [AKA statistically] predictable. Sudden changes seldom happen -- that's useful information. Trends (even nonlinear trends) continue until they don't -- that's useful information. These are natural rules of momentum -- defined in Newton's Laws. Momentum makes the future somewhat predictable. That's not opinion, that's fact.

You can "eyeball" and look for cycles (a nonlinear element in trends), and you can draw linear trend lines; and you can combine all these (nonlinear and linear) inputs and try to predict the future. This is what you are doing now. And you can backtest these methods under different market conditions and come up with statistical performance for the past which should reach, by virtue of momentum, a little ways into the future. And if the odds are in your favor, then you trade it. It is really that simple. But most people do not have the ability to do nonlinear statistical analysis in their head, nor do they have the ability to program it into a computer. But guys like you and me do have that computer-analysis ability. I would have no difficulty (and I suspect you wouldn't either) taking many intervals of historic data and determining a polynomial best-fit equation for that data; and once I have the equation, I can predict the nonlinear future.

So you do a nonlinear [for instance, cubic spline, or other polynomial] interpolation of past data then extrapolate into the future. That's your statistical prediction, which can be backtested, and can be profitable, and can be applied.