Zero sample correlation does not necessarily mean there is no linear relationship - page 24
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
if only there was something to quarrel about. Abstract questions, practically useless for the small speculator. Let the owners of large portfolios or quick access to trading platforms rack and pinion their brains. :)
The point is that these operations, to obtain statistical values (estimates), can be performed on the data series, which are suitable for this purpose. In our case, the series was not suitable, and therefore there can be no statistical estimate.
This is not an abstraction, but a very real approach.
The question is that these operations, to obtain statistical values (estimates), can be carried out on data series which are suitable for this purpose. In our case, the series was not suitable, so there can be no statistical estimate.
these discussions are empty. Whether the author has errors or not is unimportant. For this reason we must proceed from the goal - how to make profit on kotnkrete market at a specific time (even long enough to detect statistical estimates). And not to calculate something incomprehensible and try to put it somewhere.
Colleagues, it's kind of hot in here.
Везде, где читал, пишут, что нулевая корреляция выборки обозначает отсутствие линейной (обычно и слово линейная забывают) взаимосвязи на этой выборке.
They're all right. That's the way it is, and it's not an option. The only thing to note is that two very similar signals when one is shifted relative to the other (delay, shift, phase shift, ... whatever you want to call it) can give zero correlation. This is a known problem in DSP practice, and is often solved by simple overshoot.
Dickens:
No comment ...
Example of two graphs with zero MO, variance one and zero correlation. That is, the correlation in this case is the sum of the products of the BP terms divided by the length of the BP. These are EURUSD and GBPUSD charts at the interval 2010.09.28 13:45 - 2010.09.29 14:15.
Is that how you obtained the mathematical expectation of quotes? I hate to break it to you, but it's just a sampling average and has nothing to do with the MO. And surely a quote cannot have an average of zero, 2012 would have come a bit earlier :o)
PS:
This is the fundamental difference between them. Statistics doesn't exist without a terver ...
These discussions are meaningless. Whether the author has mistakes or not is unimportant. The reason is that our goal, i.e. how to make profit on a certain market at a certain period of time (long enough to reveal the statistical evaluations), should be considered. And not to calculate something incomprehensible and try to twist it somewhere.
One way to use it:
If two highly correlated financial instruments are found, it is possible to trade their "spread", or what is otherwise known as paired trading. Or even more generally, statistical arbitrage. This is a very simple strategy, based on the use of highly correlated financial instruments. In FOREX it is practically not applicable, in funds - without problems.
But even in FOREX, it is possible to build a portfolio of "majors" with dynamically changing weight coefficients, so that its equity can be predicted...
This is an empty discussion. Whether the author is wrong or not does not matter. Because we must proceed from the goal - how to make profit on a certain market in a certain time (even if it is long enough to find statistical evaluations). And not to calculate something incomprehensible and try to twist it somewhere.
One way to use it:
If two highly correlated financial instruments are found, it is possible to trade their "spread", or what is otherwise known as paired trading. Or even more generally, statistical arbitrage. This is a very simple strategy, based on the use of highly correlated financial instruments. In FOREX it is practically not applicable, in funds - without problems.
But even in FOREX it is possible to build a portfolio of "majors" with dynamically changing weighting coefficients, so that its equity could be predictable...
I traded sperds on the mmwb about 10 years ago. It used to be a theme, but when a lot of people got hooked, this business became interesting only for those who have an advantage in execution and trading costs. This is not for the small speculator, unless of course he wants to milk the DC for delays, which is a very thankless task :)
I used to trade with sperds on the MMWB about 10 years ago. It used to be a theme, but when a lot of people got hooked, this business became interesting only for those who have an advantage in execution and trading costs. This is not for the small speculators, unless of course they want to milk the DC for delays, which is a thankless task :)
I mean the real market, not DCs. Spread trading is a special case of portfolio trading. Building such a portfolio is a solvable task.
When you get to practice, you will find out that such methods are interesting when you have several billion quid of your own in management or more than an order of magnitude of someone else's.
It's true what they say. That's the way it is, with no options. It's ironclad.
Made a pointhere about the linkage.
The only thing to note is that two very similar signals when offsetting one relative to the other (delay, offset, phase shift, ... whatever you want to call it) can give zero correlation. This is a known problem in DSP practice, which is often solved by simple overshoot.
It is a realistic situation when the autocorrelation with a small lag is zero:
Is that how you got the mathematical expectation of the quotes? I hate to break it to you, but this is just a sample average, nothing to do with MO. And surely a quote cannot have an average of zero, 2012 would have come a little earlier :o)
Those who wanted to, have long since realised that we are talking about sample estimates.
Sample mean
Sample variance