I've found the anything beyond the simplest statistics to be meaningless when it comes to trading. The reason is that most statistical techniques have fairly strong assumptions about underlying distribution, which are often very difficult to confirm or measure reliably at all. I have found that using simple "sanity check", such as positive expectancy, std(trades) not too large relative to the average trade, profit / dd much larger than one over a few years, etc. Those are the sorts of things I'd look at. Just remember, that the math behind the statistics still produces a result; you have to be careful with how strong conclusions you try to draw from them.
One problem with TA is that the future is just not very predictable. I can give a pretty sure answer for what I will be doing in one day, at one week, pretty much the same, one month? Starting to get fuzzy, one year, more like a dream than fuzzy, and ten years, well, basically, forget about it. Bollinger bands are somewhat useless for absolute measurements, misapplied statistically for sure, but what about relative measurement? How best can I define the idea "expanding volatilty" in terms that my Tradestation Software can understand?? What measure is better for that than Bollinger Bands??? I really would like to know this.
Take a look at Keltner channels. They aren't perfect but I consider them a step up from BBs. Good trading, kut2k2