"Okay, I did look. And let's accept crossings aren't good once they occur."
Sorry but I wont accept that.
IMO this is where so many Chartists/Technicians get the whole 'art' of technical analysis wrong. There are no hard and fast rules on anything. There might be statistical probabilities on many aspects of price movements, but just because it is written in books, or posts by respected posters etc, does NOT make it a fact.
I'd love someone to show me some statistics on why the 50 day average crossing the 200 day average is bad, and even if they could on a range of stocks over a period of time, it does not mean in the future the same thing will happen.
Only last week we had this wonderful down sloping wedge that meant certain things were going to happen, according to some. That wedge pattern now no longer exists, as dropping below the wedge made it more like a channel. Also if anyone cared to do some research by reading the statistics from Thomas Bulkowski's,
Encyclopedia of Chart Patterns you can see that so many 'stated facts' about a lot of TA, just simply do not stand up to scrutiny.
Also a lot of Thomas Bulkowski's statistics are available on...........
http://thepatternsite.com/
I'm a big fan of Bulkowski's work as he has worked out statistically what happens, sorry did happen, in the time periods he took the statistics.
All good Chartists/Technicians know that it is just a game of probabilities, with the expected outcome only likely to happen xx% of the time. Even then, in my own experience in using TA, I've found that these probabilities need to be broken down to 'where the stock is currently located', in terms of in a bull or bear market (like Bulkowski does), but even further into where and what the underlying fundamentals are showing. Then there are other certain characteristics any one stock (or group of stocks) may show that are different in nature to other groups of stocks.
I know I'm getting in deep here, so I'll keep explaining, sorry for long post.
Do junior explorers statistically move out of patterns, MA crossovers etc the same way as miners, or retailers, or banks? In my experience the answer is NO!! When I have worked out probabilities of certain actions happening from some pattern, I've found, that different types of stocks DO NOT have the same probabilities of success, but when I break it down I find that it is also different periods of time when some things work, and others when they don't (plus the more I break up a pattern into different categories, the less data available and therefore less reliability of the resulting probability).
Following on, another aspect I've discovered over the years, is how some patterns/events seem to work all over the place (different stocks) all within a month or two of each other, then the same pattern stops working, again all over the place.
I assume that just as people start to recognize the pattern working, the hedge funds etc, deliberately start to 'paint' the pattern in the charts of stocks, then run the stops, or use it to sell into etc, to move a large position.
Again sorry to carry on so much about it, but so many people think that good Chartists with professional looking charts, signals etc must know the 'secrets' of the dark art of TA, when the reality is they don't!! It is all a series of probabilities, mostly made from unreliable statistics, that only highly experienced Chartists/Technicians eventually work out for themselves, hence why it is really an ART form, rather than a science.