Quote from bwolinsky:
Though easy, the candlestick optimizations aren't good methods, and I don't have to do any backtests to prove that to me, but as long as it goes without saying I already know that those methods are not useful and they are really the best examples of curve fitting out there. You shouldn't depend on them at all because the candlestick methods some naieve institutions use to find trading strategies do not have any predictive power. I've seen optimizations of such methods, and they do not ever produce edge beyond 65%, and if your win percentages are less than that there are far better things to devote your time to than candlestick charts and backtests. It is overly simplistic, but just as CNBC anchors point out all of the useless statistics about what percentage moves mean in terms of future price changes they don't have any sound methodology to perform anything that I would call a highly robust backtest.
If you move past the candlestick chart curve fitting methods, you'll find edges if you are creative enough with your analyses, because I already know that this is like chasing momentum and the Hershits methods tell you to do this but it doesn't work, so do not expect such methods to ever be profitable long term as I already know those particular methodologies are inherently flawed and have been shown to be unprofitable long term.
Please don't think I'm trying to be discouraging, because only a quant like mind will ever produce edge long term, and while I don't think it can be taught the mathematical aptitudes applicable to statistical analysis require much more thought than the ones you are optimizing for.
