CORRELATION EminiS&P-Russel2000

[NickBarings]

Anybody has any numbers oe an estimation on this correlation ?

 

[Jabez1947]

None that you can take to the Bank. Followed them both for years.

 

[abogdan]

Actually, I have an arbitrage system that uses this spread. You can make around 0.5 point a day fairly reliably.

 

[stock777]

Dunno, since there's a fairly flimsy corellation , how can you claim a half point is reliable. And its not arbitrage, its a pair trade if anything.

 

[abogdan]

Quote from stock777:

Dunno, since there's a fairly flimsy correlation , how can you claim a half point is reliable. And its not arbitrage, its a pair trade if anything.

If you balance the contract value precisely it becomes the arbitrage. The trick it is to find the mutually acceptable contract size that makes correlation higher than 0.95
Cheers,

 

[ProfLogic]

There is as much correlation between the ES & AB as there is between a Porsche AWD Cabriolet and a Volkswagen Bug.

 

[riskarb]

Quote from abogdan:

If you balance the contract value precisely it becomes the arbitrage. The trick it is to find the mutually acceptable contract size that makes correlation higher than 0.95
Cheers,

Absurd... but funny

 

[stock777]

Quote from riskarb:

Absurd... but funny

I was just about to devote the next 100 days to the search for the equalibrium point. Are you saying I should find a new hobby?

 

[mind]

Quote from abogdan:

If you balance the contract value precisely it becomes the arbitrage. The trick it is to find the mutually acceptable contract size that makes correlation higher than 0.95
Cheers,

i assume this indicates something like vola adjustment for position size in both instruments. i assume that "position size" or "number of contracts" is what you actually mean instead of contract size.
cointegration does a job like this. if you have two data series, you try to find a linear combination of the two that is stationary, thus fluctuates more or less regularily [i am not an english native, and "regularily" uses to remind me of that - it "nods" my tongue! even when i just write it!] around zero.
now it could be - and if that is the case i would find this interesting - that you found that by multiplying the russel with the voladifference [maybe expressed int erms of range to make it adjust immediately] of russel and emini you end up with a "new" russel that correlates highly with eminis. funny thing to think of what you are actually trading then ... probably a mixture between delta- and vegaCorrelation.

peace

 

[FredBloggs]

1 es = $12.5 per tick
1 er = $10 per tick

so you need 8 ($100) es against 10 ($100) er, or multiple of that to have a balanced spread right?


just curious