Thread: Market Neutral Corner

Market Neutral is a perplexing oversimplified term; impartial to exactly what? Neutral to certain currency vulnerability? Neutral to equity market risk overall? Neutral to Delta? Neutral to what kind of basis risk?

After we employ a makeup *cough* I mean stationary process on financial time series, sure it's de-trended enough to look impartial to the naked-eye.

Ex.
DOW DAX, Dollar Neutral Without Makeup


DOW DAX, With Eyeliner, Foundation Shade, Lipsticks, etc


Neutral to what precisely is the mega-billion question to which I do not have an answer at the moment.

Neutral to direction. .

Considering DAX-FTSE - correlation of 93.5percent (http://www.myfxbook.com/forex-market/correlation), had you purchased FTSE and bought DAX at amounts marked, and if you got out at amounts indied you could have been in profit. . So called distribute trading, nothing new, you simply need to have time period in mind and profit and loss numbers (like shutting with 5 percent of a reduction or 5 percent of a profit or following 1 week of holding). . There are no stop losses and make profits when there is only time, direction is irrelevant. . Since two different markets are involved and the volatility is different I believe 4:1 FTSE: DAX ratio is suitable if my mathematics is correct. .

Originally Posted by ;

Neutral to direction.. Considering DAX-FTSE - correlation of 93.5percent (http://www.myfxbook.com/forex-market/correlation), had you purchased FTSE and purchased DAX at levels marked, and if you have out at levels indied you could have been in profit. . So called distribute trading, nothing new, you just need to have time frame in mind and profit and loss numbers (like shutting using 5% of a reduction or 5 percent of a profit or following 1 week of holding). . There are no stop losses and take profits when trading like this, there is only time, management is irrelevant....

Neutral to management, yes. You have to get a definition for direction; unlike dollar impartial, beta impartial which describe underlying risk exposure, management neutral is vague if not subjective.

For pure Pair-trading dispersing, yes your egy could've worked with additional specialized rules. For market impartial? Correlation in direction does not guarantee co-movement in space.

Thanks

Nicely neutrality could be a lot of things..to define direction I use moving average. That's my direction. The title of this sport is risk management, not to make an omelet.

I have been reading about the beta neutral egies and like every other egy it induces a degree of complexity to a system. As long as I've understood systems have a problem with maintenance. When they start to brake down there is.

Therefore I opted to get a more simplistic approach. In a pair distributing all that I care is the level of significance as well as the volatility of pairs exchanged. In my case I first looked at FTSE and made a call that it will fall, but when I'm wrong in that call, I will supplement that with extended DAX so even when I'm wrong my losses will not be enormous in comparison to just directional commerce. On top of that the only other issue is volatility, whether something will fall or grow quicker than anything else.

Again a high correlation coefficient isn't sufficient. Neither is it!

This picture reveals two counter cases.
The two series on top have a correlation coefficient of 100%. They will never meet.
The two series at the base are cointegrated per-construction. They have a correlation coefficient of -0.02%.

Is this a real-life example or any model, because with mathematics you can prove whatever you want. . Have you ever been reading about seasonal cycles in futures. . .spreads in grains such as? Correlation works and is quite high there. .

Originally Posted by ;

Again a high correlation coefficient is not sufficient. Neither is it! This picture reveals two counter cases. The two series on top have a correlation coefficient of 100%. They won't ever meet. The two series in the base are cointegrated per-construction. They have a correlation coefficient of -0.02%. picture

Superb point. Thanks for the heads up... Very enlightening. .
Any suggestions to address this circumstance? Maybe slope correlation, or another method you have in mind...

Originally Posted by ;

Can it be a real life example or any mathematical model, because with math it is possible to prove whatever you want. . Have you ever been studying about cycles in stocks. . .spreads in grains for example? Correlation is quite high there and works. .

Merguise

Artificial or not, PipMeUp's example ought to be sufficient enough to describe the difference between correlation and cointegration in the most logical method.

Confusing? Quite so if it is the first time you listen to the phrase cointegration.

Think about it this way:

Correlation--Jim and Kim are rushing, they'all try to be the first to reach the finish line, 10,000 miles straight ahead. They all are going straight ahead, however Jim is a more driver compared to Kim. Perhaps it doesn't make a lot of gap but Jim is currently distancing himself from Kim the slow driver, along with also the distance between them could be arbitrarily long as the race progresses.

Cointegration--If they have comparable ability level however, they'll take lead in turns and also the maximum distance between 2 cars will not exceed certain amount.

Therefore you need cointegrated-correlated pairs?

Originally Posted by ;

The two show at the base are cointegrated per-construction. They have a correlation coefficient of -0.02%. picture

Take every tenth point (higher timeframe), your cointegration caliber is not going to change, but your correlation coefficient increases. I am not a mathematician, but it looks to me that the greater the tf you move to (the more you smooth out the wiggles that give you the minimal correlation) the more you'd approach the correlation coefficient of 1.
k