Everything's correlated? :(

lentus said:
Hard to disagree with any of the above. Do you find it hard to find markets with correlation close to 0? Basically something in the -10 to +10 range.
More...

There's where you will need to define your periodicity, and whatever periodicity you trade in.

So let's say that you pick a seven day periodicity.

Many platforms allow you to do this now, where you can pick something like, say, QQQ and AAPL.

Now, just off the top of your head ... should there be a natural high correlation there?

Yeah. Absolutely.

And just set the platform, when you pull up a chart of AAPL, it has a 20 day Correlation tool, and set that to QQQ.

So if you just scan briefly, over QQQ, and it's rallying, and AAPL is not, and the 20 day correlation moves to a contextually significant area ... what might many pairs traders be thinking of doing?

Right?

But it's set to 20 days ... so that defines the periodicity, not because "that's the magic periodicity or time-frame". not at all. It's just something you can slice up into 'segments' so you can get an idea of what you are looking at, and get a feel for those two instruments, that's all.
 
lentus said:
Let's say I want to play risk-off (just an example, let's not deepdive into whether that's a good idea etc). Great.

I'm long:
gold
silver
yen
swissie
long-term govt bonds of developed nations

Short:
equities
oil

Isn't all of this highly correlated and effectively just one trade? If so, how do I find uncorrelated asset classes?
More...
as a result that in bottom-line people are determined the prices and not the real value of things, products are more or less correlated (mostly in the long run and without any extreme events in certain product).
 
raVar said:
There's where you will need to define your periodicity, and whatever periodicity you trade in.

So let's say that you pick a seven day periodicity.

Many platforms allow you to do this now, where you can pick something like, say, QQQ and AAPL.

Now, just off the top of your head ... should there be a natural high correlation there?

Yeah. Absolutely.

And just set the platform, when you pull up a chart of AAPL, it has a 20 day Correlation tool, and set that to QQQ.

So if you just scan briefly, over QQQ, and it's rallying, and AAPL is not, and the 20 day correlation moves to a contextually significant area ... what might many pairs traders be thinking of doing?

Right?

But it's set to 20 days ... so that defines the periodicity, not because "that's the magic periodicity or time-frame". not at all. It's just something you can slice up into 'segments' so you can get an idea of what you are looking at, and get a feel for those two instruments, that's all.
More...

It's really weird how correlation between the same two instruments can be positive on one timeframe and negative on another. For example daily can be negative and weekly can be positive. Total WTF.
 
lentus said:
It's really weird how correlation between the same two instruments can be positive on one timeframe and negative on another. For example daily can be negative and weekly can be positive. Total WTF.
More...

Fractal nature of the markets
 
raVar said:
Fractal nature of the markets
More...

But how is it possible? Let's say on daily we have a strong negative correlation. Let's take gold and SP for example. So gold drops every day. Then SP should rise every day. So that's 7 days. A week. So after 7 days of gold dropping SP should be up. But let's say the weekly correlation is positive - that means SP will be down. How is that possible?
 
Well, you have to get terms right. Let's start there, so we're on the same page.

So on the daily, I'm assuming you mean that you have WEAK or NON-Correlation (This is NOT the same thing as Inverse Correlation, which can be another great tool, is completely separate, and needs thought about differently) between GC and SP. Ok. On the daily. For the sake of example, let's just call it ... 0.50. In your example, SP is rallying, Gold is dropping.

But you still haven't defined your periodicity of the correlation.

For example, look at GC and something like the SPX from February to May of this year.

Gold is clearing dropping. Equities are clearly rallying.

So low correlation?

NO ... well ... maybe in the LARGEST measurement. But only the largest measurement.

You still haven't defined the periodicity that you're going to measure the correlation.

But if you look at correlation between Gold and the SPX ON A 10 DAY TIME FRAME - you will note that the correlation between Gold and the SPX goes down to -0.83 correlation, all the way up to +0.66 Correlation. If you backed up the Gold SPX correlation out to the 3 month time frame, then yeah, it's going to remain very close together.

Periodicity of the measurement of the Correlation. Because there are single days, where they've gotten too far away from each other, even though the broad context maybe be two non-correlation assets ... that on a 10 day measurement of correlation, you can clearly seem them get loose, and then get tighter
 
We can go by standard terms. Correlation can go from -1 to +1. In my example, on the daily C is let's say +0.40 but on the weekly -0.72. I did define the periodicity.

Periodicity1= Daily C=+0.40
Periodicity2= Weekly. C=-0.72

raVar said:
Well, you have to get terms right. Let's start there, so we're on the same page.

So on the daily, I'm assuming you mean that you have WEAK or NON-Correlation (This is NOT the same thing as Inverse Correlation, which can be another great tool, is completely separate, and needs thought about differently) between GC and SP. Ok. On the daily. For the sake of example, let's just call it ... 0.50. In your example, SP is rallying, Gold is dropping.

But you still haven't defined your periodicity of the correlation.

For example, look at GC and something like the SPX from February to May of this year.

Gold is clearing dropping. Equities are clearly rallying.

So low correlation?

NO ... well ... maybe in the LARGEST measurement. But only the largest measurement.

You still haven't defined the periodicity that you're going to measure the correlation.

But if you look at correlation between Gold and the SPX ON A 10 DAY TIME FRAME - you will note that the correlation between Gold and the SPX goes down to -0.83 correlation, all the way up to +0.66 Correlation. If you backed up the Gold SPX correlation out to the 3 month time frame, then yeah, it's going to remain very close together.

Periodicity of the measurement of the Correlation. Because there are single days, where they've gotten too far away from each other, even though the broad context maybe be two non-correlation assets ... that on a 10 day measurement of correlation, you can clearly seem them get loose, and then get tighter
More...
 
wrbtrader said:
To answer your thread title question...

Everything is not correlated. For example, Bitcoins, Penny Stocks and BUND futures. :D

wrbtrader
More...


Are you saying those things are not correlated to the general US market, or they are not correlated to one another? Thanks!
 
Correlation is possibly the most dangerous word in trading. Combine it with volatility and you can really get yourself into some significant unforeseen trouble. I never trusted those things. I quantified them, and managed them meticulously as risk.
 
You must log in or register to reply here.
Back
Top