CORRELATION EminiS&P-Russel2000

the reason why i would keep my hand off. like outright short selling puts without additional edge. sure during the last twenty months the vola drop favoured that part of the strategy ...

what makes me think twice about it is the context here. first abogdan is to be taken serious. second we are talking about this within another kind of pairsThread ... maybe be i should widen my pairs perspective. currently i am only doing sp500 stocks in terms of pairs trading. a red zero this year. slight gain in 2003 - hardly justifying the effort taken. seeing someone pairstrading the indices and maybe merger situations as well ... hm ...


peace
 
i have known a headtrader for a desk who was trying to do the same thing with the belgian index and french index (for those not in the know, these are neighbouring countries), he added the S&P too to this kind of arb-thing from time to time. He lost about 15 million usd in a year and lost his job. As he was an arrogant asshole, I can only say he deserved it :cool:
 
if he did just that strategy, the amount of loss should be booked at least partly on the risk management of that company.

i see the point but let us be careful with the apples and the peaces. the same strategy followed by two different teams can turn out very different results. and i would guess abogdan won't get wiped out like this. curious what (s)he will have to say on the subject.


peace
 
it was not only that: he also was short vol at under 10 on the belgian options index, which probably did not help his position either, but part of his strategy was to hedge his futures in france ...
I know what you are saying though, but I have seen very bad things happen to people where wo thinking things were correlated, this was just one example.
And yes you were right about risk management, problem was he was boss over the risk manager who was just a rookie he had just left highschool.
 
Quote from NickBarings:

Anybody has any numbers oe an estimation on this correlation ?
More...
Hey Nick,
Here is your answer. You want numbers? Why don't you do your own calculations? Even if somebody actually gave you figures here, why would you trust any other sources than your own research?

Just go into Excel and figure out correlations yourself. Do it with the "CORREL" function, look it up in help if you don't know how, any kid could do it. There's lots more statistical analysis functions, like covariance, anova, F-Test, cumulative frequency histograms, fourier analysis, regression analysis, sample-based standard deviation, you name it, more than you could ever need.

Here's an example correlation calculation I just did on the quick in Excel, testing the correlation between one month of the OHLC of 5 minute-bars on 5 indices (ES, ER2, NQ, YM, DAX). In order to do that, I set a chart (in eSignal) to one time template, selected only the OHLC average (via OHLC-1MA) in the export window, exported the data into excel, and crunched the columns.

Here are the top 10 results for correlation coefficients:

DAX/ES 0.9355
YM/ES 0.9279
ES/AB 0.9123
DAX/AB 0.9121
DAX/YM 0.7783
YM/AB 0.7607
AB/NQ 0.7283
DAX/NQ 0.6529
ES/NQ 0.4829
YM/NQ 0.1873

As you can see, the ES & DAX are extremely highly correlated (so what's new), followed by ES-YM, ES-ER2 etc. On the other end of the scale (surprise, surprise) the ES-NQ correlation is rather low, and the YM-NQ correlation is almost insignificant. Incidentally, the calculations for just the actual bar closes are almost identical.

Generally speaking, I would call anything >0.8 a high correlation, and >0.9 very high. 1.0 is identical and 0 is completely un-correlated. Negative values for negative correlations (i.e. crude oil price & airlines, prosperity & Bush, etc...)

For purposes of intraday index arbitrage, spreading, trading divergences, modelling options etc, I would look where correlations are >0.9.

This is not to say that any of this data is actually significant (it's up to you to find the good stuff yourself), but now you know how to actually test various issues of all kinds (indices, stocks, currencies, commodities etc) against each other for correlation, in various timeframes, with different parameters etc. Keep quantifying, and you'll find some amazing things. And you'll no longer be dependant on ET for answers (and flames).

Enjoy!
Scientist
 

Attachments

Quote from cvds16:

... the risk manager...was just a rookie he had just left highschool.
More...

could you give the name of this company? that's just begging to be shorted...
 
Quote from damir00:

could you give the name of this company? that's just begging to be shorted...
More...
it was a private company that has been taking over in the late nineties by a belgian bank and has been totaly restructered.
 
Quote from damir00:

interesting. YM is a rather large (market cap) subset of ES - DAX is not.
More...
... damir, I noticed you deleted the "If I had to bet, I would bet against the continuation of this correlation" part of your post... :D

That would be an unusually speculative quant now, wouldn't it? :p

Seriously though, I would agree with your comment. I find this kind of correlation pretty amazing (nearly 94% in this sample), and assume it won't go on at those figures forever. However, they have been heavily correlated for quite a while now, and I have taken a few opportune trades in times where they widely diverged, particularly in recent weeks. Knowing this has been quite a quantitative advantage in my arsenal, but I won't go into details here. I also find the harsh divergences between ES & NQ these days rather notable. They used to be a lot more correlated in the past.

Warmest Regards,
-S
 
You must log in or register to reply here.
Back
Top