|
PureBytes Links
Trading Reference Links
|
Kevin, I absolutely agree - positive correlation is of utmost importance
when looking at either relative Strength Comaparative (RSC) studies, RSC
rankings, or looking for arbitrage opportunities between highly correlated
contracts.
I often find that mathematical/statistical formulae are not always suited
to charting and T/A in general. Standard Deviation is a prime example,
where it may work on nicely bell-like distributed data, but fails when
applied to market data (apologies to Bollinger Bands fans).
Through much testing I'm finding that traditional correlation methods may
not work so well on the markets either. Pearson's correlation (+1/0/-1)
may fall into this category, as correlation between securities is hardly
ever linear (i.e., dynamic).
Spearman's correlation looks more promising, but is probably not doable
using plain MetaStock Formula Language, as it seems to involve ranking.
http://en.wikipedia.org/wiki/Spearman's_rank_correlation_coefficient
Jose's upcoming correlation in the URSC toolkit is looking promising, but
then again I may be biased here... ;)
jose '-)
http://www.metastocktools.com
--- In equismetastock@xxxxxxxxxxxxxxx, Kevin Barry <kevin_barry@xxx>
wrote:
>
> Hello Jose,
>
> Very interesting to hear that you are looking at this subject.
> I think that the pairs trading technique is very under appreciated
> probably because of the difficulty in setting it up properly in the first
> place.
>
> The fist step is to find correlated securities, or sister stocks as
> Jesse Livermore describes them. Unfortunately, (it appears to me that)
> Metastock is not a suitable tool to do the job. In effect, we need to
> compare the potential correlation of securities using some kind of
> matrix system.
> Excel would appear to be a better tool and it already has the Spearman
> and Pearson functions built in which is probably a good place to start.
>
> Once correlated securities have been identified, perhaps Metastock then
> becomes more useful. We then need to scan our shortlist for occurrences
> of deviation in order to generate our trading signals. I have been
> playing around with setting up a unique indicator for each pair of
> securities.
> To date, I have been scanning the resulting charts manually hence my
> original question to the group about making the visual representation
> easier to read.
>
> Looking forward to developments!
>
> Regards,
> Kevin
>
>
> At 16:16 15/04/2007 -0700, you wrote:
>
> Andrew, I agree that correlation is a huge subject.
> MetaStock's hindsight-based measurement of correlation doesn't cut it.
>
> I've been working on just about every kind of correlation feasible in
> MetaStock, and so far I've found that measuring correlation between two
> securities based on the following, is the most promising:
>
> 1) a properly constructed Rate of Change % (i.e. not MS's RoC), and
> thus deviance % from each security;
>
> 2) or base correlation on directional changes - that is, both securities'
> direction (measured to a common lookback starting point) up/equal/down
> for positive correlation, opposite directions for negative correlation.
>
> So far I've found method #2 above the most promising candidate, as long
> as the RoC's lookback period is based on time (calendar days) rather than
> possibly irregularly spaced bars, and the equal zone has some % leeway
> either side of it.
>
> I may spend some more time on Log(RoC%) and see if there is anything
> viable there.
>
> Any thoughts on this matter appreciated.
>
> jose '-)
> http://www.metastocktools.com
>
>
>
>--- In equismetastock@xxxxxxxxxxxxxxx, andrew_tomlinson@ wrote:
>
> Also you need to consider whether you are correlating raw prices, %
>changes, log changes, etc. Makes abig difference.
>
> Andrew
>
>
>
> -------------- Original message --------------
> From: "Jose Silva" <josesilva22@>
>
> It's funny you should ask about this, Kevin, because I'm in the middle
> of incorporating something similar into the next major release of the
> URSC toolkit.
>
> Basically, what you are asking is for a comparison of the two contracts'
> rate of change. There are many issues involved in plotting a valid
> comparison (such as a synchronized starting point and a valid % RoC) and
> the URSC toolkit covers all the important ones.
>
> More here:
>
> http://www.metastocktools.com/URSC/URSC.htm
>
>
> jose '-)
> http://www.metastocktools.com
>
>
>
> --- In equismetastock@xxxxxxxxxxxxxxx, Kevin Barry <kevin_barry@>
> wrote:
>
> Hello,
>
> Could I pick your (collective) brains?
>
> Two securities can be compared using the Security function and then
> subtracting or dividing one plot from the other. The results in a
> trending plot.
>
> What I would like to achieve is to plot equity A as a straight line and
> then have the second plot displaying the difference between equity A and
> equity B. Ideally, this plot would indicate the percentage deviation
> rather than just the number of points difference.
>
> For those of you interested, the idea behind Pairs Trading is to
> identify two stocks that are closely correlated. When a divergence is
> detected, one would go long, say, equity A and short equity B in
> anticipation of the correlation being reestablished in the future.
> It is a market neutral strategy.
>
> Is this doable in Metastock?
>
> Regards,
> Kevin
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/equismetastock/
<*> Your email settings:
Individual Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/equismetastock/join
(Yahoo! ID required)
<*> To change settings via email:
mailto:equismetastock-digest@xxxxxxxxxxxxxxx
mailto:equismetastock-fullfeatured@xxxxxxxxxxxxxxx
<*> To unsubscribe from this group, send an email to:
equismetastock-unsubscribe@xxxxxxxxxxxxxxx
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
|