Technical analysis works if the majority of the people use roughly the same things. If the majority of people think fib numbers work then they may work. If no one on earth besides me looks at a 34EMA then I doubt a 34EMA would be of much use. Trading with the trend means trading with the herd. http://tradingcrude.blogspot.com/
I really don't want this thread to be about my trading, but I will answer your question. When there is a H/L or L/H failure, I simply wait for the next structure that presents itself. As you say, it is about "odds" or tendency. However, I don't preordain where those highs or lows will be (as Fib users appear to do). I react to them. And, as no doubt you have gathered by now, I am not a fan or prediction or preordination. That's just a tad too mystical for my taste.
If that is so, then its effect will be reflected in the price action (on those occasions that it does work) and you will be able to react accordingly. On the occasions that it does not work, price will not reflect it, and you will be able to safely ignore it. Effectively, then, you can ignore it altogether because price will apprise you when to act. And that is the definition of superfluous. Unless, of course, you are planning to predict the future with those magic beans.
I don't know why (or how) someone would try to prove Fibs "work" from a statistical perspective. I'd be curious what formulas they used to construct their tests. I would think any statistical test for any commonly used indicator would fail... they have to fail or else they would cease to be useful (unless we're talking about lagging indicators use in a lagging way, which are never useful IMO). I personally find Fib retracements to be an excellent tool for discretionary trading (and yes, I do trade profitably). If you trade with a trend-following bent, depending on the price action of the last move, you may want to join that trend during a pullback. You have to pick a spot to enter, right, so why not use a fib level? Which fib level you choose could be variable based on the volatility of the last move. Does the market have to bounce off that level to the tick for you to say that fibs "work?" If you want to be long, find a good spot to get long, define your risk, and make yourself available to the opportunity you perceive.
Removing astrological component would maybe please Occam, but it would also take away any 'order' to why price move like it does...
Yes, it would certainly remove preordained order. Are you trying to predict price like a god or just trade it like a mortal? Are you trading what you see or what you "believe?" Are you accepting the market structure as it presents itself, or are you trying to impose your own order onto it?
Forget those complex E. Wave, Fib, etc. Back to basic, what Nicolas Darvas had found is still true. The attractive stocks continue to have the same characteristic. If a tempestuous beauty were to jump on a table and do a wild dance, no one would be particularly astonished. That is the sort of characteristic behavior people have come to expect from her. But if a dignified matron were to suddenly do the same, this would be unusual and people would immediately say, "There is something strange here - something has happened." A stock that continues to rise significantly after the open for two or more consecutive days under the circumstances of Bear Market seem to be unusual and therefore attractive.
I did not mean 'preordained' order, I mean 'order' in the here and now... We all trade our believes and see what we believe. Not the other way around...