Bollinger Bands: Calculation flawed ?

BSD

BSD

Veteren member
Messages
3,819
Likes
988
Found this via TraderMike's blog. Don't use BB's myself and on top of that I'm not a numbers crunching quant so really can't comment, just found it interesting that they've been around for ages and nobody else has ever complained ?

"OBSERVATIONS AND COMMENTS
Bollinger Bands and Hamzei Sigma Channels suffer from the same type of flaws as the Kase Dev-stops. To repeat, there are obvious procedural problems in terms of how the numbers are handled. I could go into details, but it’s not my mission to fix other people’s products. Suffice to say that the numbers are wrong. Better yet, I will prove it."

Also check out her general thoughts on technical indicators:

"First of all, I am only familiar with technical indicators in terms of how they can be represented graphically; that is, I plot the information as lines, dots, histograms and color bars in ways that make intuitive sense. A familiar look and feel makes the tools easy to use. And that is where the similarities end.
What goes into the calculations is another matter. You might be surprised, but I actually know very little about technical indicators. Why? Because most of the work would not pass muster due to conceptual and procedural flaws. If you want to kill a quant with laughter, just show him a bunch of canned equations from a charting program."

I'm no big friend of indicators myself, preferring to keep it simple with not much more than some moving averages to pinpoint trend, and that apart using price itself as my main indicator, but I'd still find it interesting to say the least if her allegations were to be true.
 
All technical indicators are flawed IMO. Furthermore, i think all TI's are roughly equal to each other in their overall usefulness. I can't name one TI that stands out head & shoulders above the rest.
 
  • Like
Reactions: BSD
GammaJammer said:
Did she expand at all on the specific flaws she perceives in BB methodology? I often have them up on my charts but more as a quick and dirty measure of volatility than anything else.

GJ
Click to expand...

Hi Gamma, agree that it's all a bit thin, all I found was this from the second link above at the bottom of the page:

"I was disappointed to find that the work was not done according to standard exploratory data analysis procedures. Example: any time we measure something, it should be done in units that allow apples-to-apples comparisons. Another example: the standard deviation of a probability distribution known to be non-normal is of little use.
If we subject similar indicators such as Bollinger Bands and Hamzei Sigma Channels to the same rigorous tests, they would also have difficulty making the grade."

PS: link in the quote is from her.
 
GammaJammer said:
Is that an inadvertant pun there JT?
Click to expand...

Sorry mate, i don't understand :confused: .
I just mean, all TI's work well sometimes, and become completely useless at other times - BB's, RSI, donchian channels, MACD etc. all fit into this category. You only need to put them on a chart to recognise this inconsistency/unreliability. But used correctly, with the right parameters, timeframe etc. they can still turn over a profit.
 
Don't wan't to answer my own question here as I really don't know, but just googled a bit as I'm really quite intrigued by now and came up with this:

"The idea of Bollinger Bands is flawed mathematically, but the reason is complicated. I wouldn't use them, but here is the idea.

There are several measures of statistical distributions. The four primary ones are mean, standard deviation, skewness and kurtosis presuming your "residuals" are normally distributed.

Skew is the measure of shift of the median from the mean, or how squished to one side the actual distribution is. Kurtosis is the measure of the fatness of tails.

Bollinger Bands ignore skew and kurtosis, the result is, they do not work.

The mean is what is usually called the "average." Standard deviation is the square root of what is usually called the variance, or how wiggly the data is around the mean.

Stock prices vary, so Bollinger Bands look at the recent mean, say of the last twenty minutes and how wiggly the price has been. Each trade the mean and variance are calculated and recalculated. As the variance increases, the bands get wider, likewise, they shift around the mean as it moves.

The basic idea of the Bollinger Band is that when a stock moves strongly above a point several standard deviations up or down, it should revert back to the rolling mean. So if a stock falls dramatically below the bottom Band, you should buy and if it shoots way above a Band, you should sell.

The weakness of the Bands are three-fold:

1) It ignores kurtosis and skew, in doing so, it is actually ignoring the true data and placing a false model on it.

2)It assumes there is sufficient information in the stock prices to make choices without additional information, however a review of the "residuals," would show that significant and important information is being ignored, forcing faulty decisions to occur.

3) Equity prices are not normally distributed and are either Cauchy or Levy-Alpha Skew distributed making point estimates meaningless using normal calculation methods."

Hmm...
 
JTrader said:
Sorry mate, i don't understand :confused: .
I just mean, all TI's work well sometimes, and become completely useless at other times - BB's, RSI, donchian channels, MACD etc. all fit into this category. You only need to put them on a chart to recognise this inconsistency/unreliability. But used correctly, with the right parameters, timeframe etc. they can still turn over a profit.
Click to expand...

As you say, they can do, but how do you figure out the correct averages to use? The whole thing about indicators, to me, is that they are a consequence of the chart pattern that has appeared, in the first place. I never have been able to figure out the rationale behind this indicator idea. Like the divergence. That seems to work very well on the extreme low, when the indicator shows a higher low. I've used that many times with success. Using it on the tops, though, quite often, when the indicator shows a divergence, the chart keeps on going up for some time before it tops out, finally.

Split
 
So Yahoo had this to say:

"The idea of Bollinger Bands is flawed mathematically, but the reason is complicated.

The weakness of the Bands are three-fold:

1) It ignores kurtosis and skew, in doing so, it is actually ignoring the true data and placing a false model on it.

2)It assumes there is sufficient information in the stock prices to make choices without additional information, however a review of the "residuals," would show that significant and important information is being ignored, forcing faulty decisions to occur.

3) Equity prices are not normally distributed and are either Cauchy or Levy-Alpha Skew distributed making point estimates meaningless using normal calculation methods."

But at least they seem to offer self-fulfilling prohecies ?

"The bands cannot, as some have supposed, be used to make reliable statements regarding what fraction of an equity's prices will lie within a certain distance of the mean value. This is because an individual equity's price does not obey known distribution functions (see stochastic process). For example, if the bands for plus or minus two standard deviations (2SD) are computed, it is wrong to suppose that ~95% of an equity's closing prices will, on average, lie within the Bollinger bands. That would require, among other things, that the prices be normally distributed, which they are generally not. It would further require that the true standard deviation be known. The standard deviation calculated as above, however, is only an uncertain estimate of the true standard deviation.

It is of interest to note that faulty interpretation of a price touching or breaching a band based on incorrect statistical assumptions has become so widespread that some traders now use these events alone as trading signals and by so doing may have unwittingly injected significance into these band-touching events that should otherwise be absent."

Interesting. Monsieur Bollinger should have a say...
 
I read the article in question too and really got very little out of it. As has been noted, it's short on specifics, especially when the author starts talking about her better indicators.

She's definitely got a point on the poor calculation methodology of the Bands, but only if you expect to use them to provide an expectation of future prices. If that's the case, you should be using returns, not prices. My own use of the Bands is strictly as a volatility reading, and to that end just ploting the Standard Deviation of closing price serves the same purpose.

The larger point of about technical indicators being flawed is a legitimate one. If you showed the vast majority of them to statisticians and mathematicians they would laugh. I'm certainly not in that group myself, but I do have enough quantitative training to be extremely dubious about most of it all. Moving averages? Come on. The fact that price is above the average only means one thing - that the average is going to rise.
 
  • Like
Reactions: BSD
In my view it depends on what you intend to use Bollinger Bands for as there are several ways of using them some of which can be quite effective.


Paul
 
GammaJammer said:
Many people prefer to trade continuation, and view a full bar close outside the band as a strong indication of a trending market.
Click to expand...

That's definitely me, keeping it simple and just buying when things move up, or shorting when they move down ;-)

Matter of fact I believe awhile ago there was some mechanical system called Aberration or something that did just what you described across lots of markets, and some people made millions of off that apparently, I think Mark Johnson who posted a lot on some forums awhile back was one of them. He had a mention in Curtis Faith's book, "Way of the Turtle".
 
JTrader said:
Sorry mate, i don't understand :confused: .
I just mean, all TI's work well sometimes, and become completely useless at other times - BB's, RSI, donchian channels, MACD etc. all fit into this category. You only need to put them on a chart to recognise this inconsistency/unreliability. But used correctly, with the right parameters, timeframe etc. they can still turn over a profit.
Click to expand...

This work/not work business has to do with a fundamental misunderstanding (a general misunderstanding, not necessarily yours) of how "indicators" are calculated and what they're supposed to do. We think we see the indicators indicating something, or not, and believe we have made an important discovery. We then devote our efforts to improving the hit rate and the probability of whatever it is we think the indicator is indicating when our efforts ought to be focused on detemining whether or not the indicator is actually indicating what we think it's indicating. In most (all?) cases, it isn't.

Consider the virgin being tossed into the volcano: sometimes it results in a great crop, sometimes it doesn't. Maybe tossing her in earlier or later will change the probability of a healthy crop. Maybe two virgins are better than one. Maybe six. Maybe tall virgins are more effective than short ones. And surely age is important. But does the robustness of the crop really have anything to do with tossing the virgin into the volcano in the first place?

Db
 
  • Like
Reactions: BSD
GammaJammer said:
As I understand it this is a bit of a popular misconception. Many people prefer to trade continuation, and view a full bar close outside the band as a strong indication of a trending market.

Could be wrong, but that's what I thought. But you're right - lots of people bet the ranch on mean reversion all the time. Sometimes it works, sometimes it doesn't ;)
Click to expand...

Bollinger himself actually made a similar point.
Bollinger Bands - The Basic Rules
 
GammaJammer said:
In particular;



GJ
Click to expand...
I rather like BBnds, but would never suggest using them alone as a signal generating tool. I use them with different deviation settings, either standatd or fibs, on different timeframes. If pressed I could compare it to motorway driving, one does not join or exit the motorway to/or from the extreeme right lane (fast lane???), but many use it to travel on it. Does it mean the 3rd or 4th lane is bad/wrong? If one is on it it is at least safe/profitable to be aware of it as a fact. Similar with the price, BBnds provides for me a qiuck indication as to the speed of the price change in relation to time and the past (yes historic) performance. Books like "Bollinger on Bollinger bands" by John Bollinger, or "Technical Analysis of the currency Market" by Boris Schlossenberg have been helpfull to me in the past. BBnds like all the rest of the indicators would only indicate what the bars, or candles are telling us anyway, but as a shothcut they are usefull for providing an orientation at a single glance at the chart.
 
Last edited:
BSD said:
"OBSERVATIONS AND COMMENTS
Bollinger Bands and Hamzei Sigma Channels suffer from the same type of flaws as the Kase Dev-stops. To repeat, there are obvious procedural problems in terms of how the numbers are handled. I could go into details, but it’s not my mission to fix other people’s products. Suffice to say that the numbers are wrong. Better yet, I will prove it."
Click to expand...

From a true statistical perspective the basis of bollinger bands are flawed, but that doesn't mean that they aren't a useful tool.

I wouldnt get too hung up on the application of rigorous statistical techniques. After all I believe a statistician once drowned in a lake averaging only 2 inches in depth
 
zupcon said:
I wouldnt get too hung up on the application of rigorous statistical techniques. After all I believe a statistician once drowned in a lake averaging only 2 inches in depth
Click to expand...

LOL !

Nice one
 
A few points

1) Anyone basing buy/sell decisions using only BB's deserves having their a/c emptied. :devilish:
2) The 20/2 setting is not the only setting. ;)
3) BB's together with S/R, Channel Analysis, Fibs, Pivots, Candle Analysis, Momentum Analysis etc is a valid way of using BB's.
 
  • Like
Reactions: BSD
fibonelli said:
A few points

1) Anyone basing buy/sell decisions using only BB's deserves having their a/c emptied. :devilish:
2) The 20/2 setting is not the only setting. ;)
3) BB's together with S/R, Channel Analysis, Fibs, Pivots, Candle Analysis, Momentum Analysis etc is a valid way of using BB's.
Click to expand...

Makes sense
grinning-smiley-003.gif
 
You must log in or register to reply here.

Similar threads

M
Bollinger Bands
2
Replies
20
Views
5K
rathcoole_exile
rathcoole_exile
John Bollinger
  • Article
Technical Analysis Bollinger Bands – The Basic Rules
Replies
15
Views
31K
StkWinGuy
S
M
What Are Bollinger Bands?
Replies
13
Views
3K
Lightning McQueen
Lightning McQueen
A
Bollinger band excel calculation?
Replies
12
Views
35K
cowbearcar
C
C
Help Understanding Bollinger Bands
Replies
0
Views
2K
Cadavre
C
Top