> Basically what your saying is: It works, if by chance you pick the right model out of 'n' models, with 'n' approaching infinity.
No, that's actually not what he's saying at all.
If I'd get the impression that you are at least ever so slightly inclined to at least try to learn how it might work, I'd give a slightly more formal explanation of what (I believe) it does, and what it doesn't do.
But as it stands now, judging by the style of your answer, I'd waste my breath (or rather: my time, typing out a longer reply)
> Again, if Elliott Wave theory would work you would have an algorithm to do the prediction job for you.
Got it. If it's not fully algorithmic, it doesn't work.
I made this analogy before, in a different discussion, but here it goes again: Tell that ("if it's not formal enough to be an algorithm/computable, it's worthless") to the early computer scientists programming chess engines with insufficient processing power to brute force a victory against a human GM (basically, the state of the world until IBM came along and set their goals). A lot of what a human GM does, apart from the inate intelligence and year after year of training is learning the theory of the game. If you'd ever happen to look into a book of, say, opening theory, you'd notice pretty quickly that the strategies outlined in there are decidedly "unalgorithmic". Didn't change the fact that those who learned the strategies were better than a purely algorithmic machine (until Moore's law took care of that, and the implementation of the theoretic aspects of chess *did* in fact progress far enough, more recently).
tl;dr If it's not algorithmic, it's difficult to show (in an isolated experiment) that it works, but from "difficult to show in isolation that it works" you shouldn't conclude that it doesn't work.
The reason why EW analysis is done by humans is not that an algorithm would be (at first!) strategically inefficient (as in the chess example). The reason is that the starting parameters are rather arbitrary, because the EW rule set does not define them properly.
But as logic suggests, if the propositions of a statement are arbitrary then the possible conclusions are also arbitrary. So EW theory does not predict anything. It only shows arbitrary possibilities.
ya.ya.yo!
You're inching towards an improved understanding :)
1) It provides different parses of the trade data. More specifically, EW is used to: (a) (more or less) exhaustively show all possible parses (of the data, into trends), and (b) to allow the trader to list them non-redundantly. Which all by itself is already a huge help for the human brain, to look at the historic data and consider the different ways the data can be broken down into trends. (EDIT: "helpful", because it is now possible to do so without running into loops)
2) EW then suggests likely continuations, given a parse, and invalidation targets. These are essentially (in my opinion, at least) based on simple market insights/heuristics. Think heuristics along the line of "a series of lower lows forms a downtrend".
The point of those heuristics (apart from enabling the exhaustive parsing in step 1) is to list (and maybe rank, by likelihood) what in the past (across different markets) seemed to have been commonly encountered continuations given a parse of the historic data. Which means any suggested continuation is not necessarily the "right" one, and neither is any parse of the historic data. However, as long as there is *something* correctly captured about the conditional probabilities of the market in the suggested continuations, then step 2) is useful as well.
