Bollinger band question

[dgproject]

I am currently doing some TA-research and would like to backtest Bollinger Band signals. For the upper band, I have to add 2 * standard deviation to the MA-level.
However: what kind of sd do I need? Is it
(1) as often suggested the sd of the difference between the MA and the closing price? or
(2) the standard deviation of the share price itself. This procedure seems to be used by the TA-lib excel add in.

(and if (1) is the correct version, do you guys have any experience with reliability of the TA-lib formulas?)

Many thanks in advance!

 

[Rhody Trader]

Standard application of Bollinger Bands uses the standard deviation of closing price.

 

[dgproject]

Standard application of Bollinger Bands uses the standard deviation of closing price.

Thanks for your swift reply!

 

[arul57]

yup agree with rhody, just use the BB standard...

 

[Splitlink]

I believe that you should use more than the 2 SDs because economists use populations of hundreds and thousands for accurate reading. Most of us are using averages. How big is an average in TA? Most times hardly more 30-40 , probably. That distorts the accuracy.

All that said, I get some good signals off of BBs and they are the only indicator that I use, apart from averages because they do give overbought/oversold signals, which I find useful. One has to decide, from reading the bars on lower TFs, IMO, how to trade this signal because it could be a bounce, or the start of a trend.

Remember the cons and you should find the pros useful.

In a trending market, let's say down, The piercing of the upper band happens more rarely, the price staying mainly in the area below the average and the lower Bollinger which indicates that the average, itself, should be the target for entering.

Nothing written in stone, here! Just an opinion.

 

[Splitlink]

As to which SD to use, I would suggest looking periodically at the chart over a year and letting a good, old, eye-ball tell what works best for you because there are a lot of opinions. Your own is what you should work with if it gets good results.

 

[Rhody Trader]

I believe that you should use more than the 2 SDs because economists use populations of hundreds and thousands for accurate reading. Most of us are using averages. How big is an average in TA? Most times hardly more 30-40 , probably. That distorts the accuracy.

The argument I would make here is that we're talking about closing price for the Bands, which is not anything like normally distributed (period % change would be better, though technically that isn't truly normally distributed either, though it's a lot closer). As such, you can't really make the same kind of statistical analysis as you would for other types of data. In other words, you can't use the SD of closing price to suggest that roughly 2/3rds of the time it falls within 1 devation.

Basically, all the Bands are telling you is how volatile closing price has been over X periods. What that X should be is probably best determined by the time frame in which you favor trading.

 

[Splitlink]

In other words, you can't use the SD of closing price to suggest that roughly 2/3rds of the time it falls within 1 devation.

Basically, all the Bands are telling you is how volatile closing price has been over X periods. What that X should be is probably best determined by the time frame in which you favor trading.

Don't get too technical with me, Rhody Trader, I'm, really dead ignerent if you dig too deep.

The x, for me, that you mention, is formed on the hourly chart and I use that as a target for "overbought/oversold" To get turning points I look for contractions beyond the band that I have to identify as a bounce or a continuation. I do this on a lower TF for earlier entries. I get it wrong the same as everyone else but, since it is overbought/oversold I consider the probabilily that the price will drift towards the average and I, also, try to consider the probability of getting stopped before the turn happens.

I was trying to express that, in my experience, 2 deviations has not been enough.

Always willing and happy to be put right.

Nothing too heavy, though. I left school at 16.