Howard,
I am having the same problems you are. My original concern was that most of the winners with the Bollinger Band system are small and a few large losses can dramatically reduce results. I haven't actually tested it but I suspect thats why results last year were small. A low volatility market, small profits and a few big losers. My thought is to either add a filter to improve winning percentage or find a way to achieve larger wins. With that thought in mind, my first attempt at improvement I posted under the mean regression thread. Maybe you wouldn't mind taking a look and see if you come up with the same results.
I'm sure Larry will come up with a better idea(s) or maybe the system is just fine the way it is and doesn't need tweaking and I'm wasting my time
Jim