|
PureBytes Links
Trading Reference Links
|
Hi Brent
----------
> From: BrentinUtahsDixie <brente@xxxxxxxxxxxx>
> To: RealTraders Discussion Group <realtraders@xxxxxxxxxxxxxx>
> Subject: Gen: Standard Deviation help.
> Date: Thursday, June 11, 1998 3:57 PM
>
> RT's,
>
> I have one of those Statistics 101 questions(which I never took) about
the
> use and calculation of Standard Deviation. In my SuperCharts software
there
> is a StdDev function, it allows you to input variables for price and
> length. I have checked out the definition of Standard Deviation and I
know
> it’s something like the Square Root of the variance. Now I have seen or
> read many times where the designation for Standard Deviations is a set
> amount in relationship to something like price such as, 1 Standard
> Deviation, 2 Standard Deviations, 3 Standard Deviations... what I don't
get
> is, is how this set amount is derived. Now if it is just the value of the
> last calculation(after a series of calculations) extended back in time
I'd
> like to know that, otherwise can anyone tell me how it works. Thanks.
Standard deviation is calculated as follows on a given set of data:
1 work out the average of the data.
e.g. 111, 114, 117, 118, 120 has an average of 116
2 deduct difference of each reading from the mean
keeping the +/- sign.
e.g. on the data above the differences are:
-5,-2,1,2,4
3 we can't take the mean of the deviations as they will always add
up to 0. To overcome this problem, square the deviations to get rid of the
minus sign.
e.g.
25,4,1,4,16
Now get the mean = 50/5=10; this mean is called the variance.
4 so that we can relate the variance to the same units of the data set,
we take the square root of the sum of the variance and that's the standard
deviation.
e.g.
in this case sq.rt of sum of variance =20.
To use this we need to be aware that if the data set is a bell curve,
then 68% the data set will achieve values of (mean +1 std). In this case,
116 +/- 20= 68%
mean +2 is achieved 95.0%
mean +3 is achieved 99.7%.
How to use in your trading? Well let's say that in a 1 period monthly
swing, the USD/DEM has a mean of 2533 points 75 days with a
std deviation of 1624. Once a mkt move achieves
mean +1 points, we know that the probability of the mkt continuing is 32%
i.e. 100-68. We thus have an objective measurement of overbought
rather than an arbitrary figure e.g. 14 days for an oscillator.
That's a brief and very crude example of the concept's applicability.
If you are interested in stats, I recommend "Statistics Without Tears" by
Derek Rowntree. Because of a misspent youth, I consider myself a
maths dunce. If Derek can teach me Stats, anyone can! Even better
the paperback should cost no more than US$10/$12
regards
ray
R Barros
101/25 Market Street
Sydney NSW 2000
Australia
Voice: 61 2 92673470
Fax: 61 2 92673478
E-Mail: rbarros@xxxxxxxxxxxxxxxxxx
|