Seems except that website, no one else has done it yet, I have to do it by my own then. Maybe pattern recognition programing is still too difficult for the most to understand, then there seems an edge there: More people can do, less profitable. Human's pattern recognition mainly based on fuzzy logic, permitting certain kind of deviation, so 20%to 50% deviation is allowed, basically pattern definition can be several speed/movement combination, I will try to program and publish my pattern recognition system later Actually I think human are quite simple, every human behavior is easy to define.
If you're thinking this way, you're doing well. The financial storm has increased volatility. Breakout methods work better during periods of exceptional volatility. There remain, however, the subjective decisions of stoploss, and profit-taking. Decreasing volatility does not necessarily mean a lack of sustainability. It means profits might not accrue as quickly as the speed of price action decreases. In other words, it will take longer for each trade to bear fruit. Trading is an art...anyone who thinks otherwise is deluded or has something to sell.
A very interesting topic is about discretionary trading and art. In order to try this, I spend almost 1 year to trade discretionary after 2 years of systematic trading. Those trades only based on my "experience" and "feeling" of the market, after several hundred of such kind of trade, I think it is the time to define those judgements in more general terms. I mean, you should have SAME judgement based on exactly the same information, otherwise the chance of being correct is near coin casting. So, if for each price action or several price movement combination, there is an fixed judgement, then it is definable. But for an artist, for the same information, there will be differenct judgement, based on the mood of the artist and maybe some other variables. But I still think this way of thinking will lead to random walk theory, if we are not seeking for certainty, what is the meaning of being in the game?
A book that I found informative on the subject was" Technical Analysis: The Complete Resource for Financial Market Technicians" by Kirkpatrick and Dahlquist. They go over the formations of the patterns and how they work out. Another web site I had found useful is: http://thepatternsite.com/visualcpindex.html (oops just noticed it was posted earlier in thread, sorry)
NihabaAshi's car analogy was spot on. The reason there will never be any real scientific studies for this is that while someone may lend someone else the tires to test, no one is going to give away the exact blue prints to the car. Without the blueprints your not really testing anything.
Combining trash with other trash does not offer any value add. A pattern/indicator/edge either works or it doesn't. Combining 2 random things only gives you more randomness.
If we can count my dragon pattern as a classic, yes I have been forward testing and using it and it has been working, thank you very much. For more info do a search on ET... The problem is with your questiuon that it puts a large and different kind of dataset together and generalizing about it...