Boy oh Boy. Bollinger bands are great. When you are in a trading range, they bounce right off the band an you can fade the trade. When you are in a trending market, you just continue going with the direction of of price action. Okay - so what are we saying here? You first have to figure out what kind of market you are in; trending or trading range? Once you know that you know whether to go with the trend or fade it. Simple isn't it? Bollinger Bands is just another tool. It is not the holy grail. You really do need to use it with other stuff. I agree with toc - you can use ADX or DMI. You can also use higher timeframes and see how the price action fits in the bigger picture. Trading isn't simple and one indicator isn't going to make you successful. Bollinger bands can be very useful - just try to fit it in with other stuff.

By the way, other than seeking 20-25 readings on the DMI, are there any other deeper insights into ADX-DMI /strategies. Trade percentage goes higher into winner coloum using these but then number of trades is also reduced. Many a times huge trades are missed because reading is like at 40+.

Been a long time since stats but the standard deviation and bell curve idea w/ 95% being within 3 standard dev's....that would be based on the mean....correct? IE in a sample, 95% are within 3 S.D's of the mean. Price in markets though, is different. Looking back, price would fall within that range, but each new tick changes the mean, the views of market players, etc. So like somebody said, you have to look a t bands like a tool. Price distribution is gonna be different than grades on a test so look at it w/ that in mind.

It does sound simple. But basically ADX is just an indicator checks for overlapping price actions so when you are in a range, it does not advance much and when it breaks out of a range, it advances. So this is just basically TRADING HINDSIGHT by using it to determine whether the price is in a range or trending when applying Bollinger Band. By that time, you probably already have your trading loss. The common mistakes (widespread) of Bollinger Band is assigning statistical probability on the standard deviation when in fact Mr. Bollinger might have made a fundamental mistake from the start by applying it on the price series instead of returns (He would get a C in Statistics 101). It is just very dangerous to listen to people's mathematical mumbo jumbo without checking it out oneself. As other people has pointed out that it can be used as an aid to track the cyclical nature of the volatility (Bollinger squeeze). But I found volatility break out not to be as profitable (easily traded) as I think... but I have to do more research on this.

Exactly. Nothing is 100% in this game. Bollinger talks about M and W formations. Bollinger Bands as with other indicators are just that, indicators. They need be used as part of the big picture.

Hi John, I was in the same spot as you until I research and refresh my Statistics 101 memory a little. I am gonna rephrase what I've already said: 1) A segment of a price series is just weird to be considered as a population. The returns of a stock in a given time interval make a better population for statistical analysis. 2) Most likely neither one of them are normally distributed since the population of returns are found to be fat-tailed. The probability assignments won't come out right. 3) Standard Bollinger band use 20 as the sample size to estimate the population standard deviation with the sample standard deviation. That is a little too low. So if you assign the bell curve probability on the price series as every freaking websites and books do... it's just doesn't make much sense. It does measure price dispersion from the mean but you just cannot assign normal population probabilities on top of the statistics. Hope it helps. I am not a math guy myself so I might not be correct 100%.

innnnteresting. Good post. Personally, i've never used bollinger bands w/ stuff I trade....I had one system that used them alot but haven't gotten it to the tradeable point yet. But I do like the 'type' of study they are...i.e. some type of average and some type of channel (to be ridiculously general). I do use bollinger bands w/ 1.8 or 2.0 standard devs on ADD, TRIN, and TICK linecharts though. I set the charts to 15 min's usually w/ a 210 period MA (roughly equal to the 5 day average). This paints a pretty good picture of when breadth is getting overextended. Although i haven't tested the idea, I find that an oversold TRIN seems to happen right before a rally and an overextended TICK right before a top. Sounds like the same thing but it's actually not.

^^^^^^^^^^^^ Like John Bollinger, he says stuff like ,stocks are like sheep; if you want some going north, find a group of stocks going north. Not a prediction. Also ,BB are not predicters of anything either, for example on a 25 year monthly chart of GE; you may find about 7 years of monthly closes outside BB. Also, 7 years of no commisssions/ slippage maybe unrealsitic also.