For those who have (are) seriously engaged in co-movement analysis of financial instruments (> 2) - page 23
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
more, let's get something concrete in the studio. Let's have a look together.
Yes, for example, what's wrong with such a synthetic:
Well, let's say people build some supersynthetic, the question is still how long this synthetic channel will last ...
Yes, for example, what's wrong with this kind of synthetic:
Off-topic. Channels like this are discussed here, for example.
Just on topic, because you want to get relatively predictable behaviour from a synthetic.
No matter how sophisticated this synthetic is, it will still eventually build on the majors,
and the majors behave the way they do, so I don't think the complexity of the synthetic will make it any more predictable.
The name of the thread is still relevant.
It discusses analysis of the co-movement of financial instruments (> 2). Synthetics are a special case of this analysis, which is why it is discussed here.
Your example does not, unfortunately, fit the topic. The name of the topic is purposely given so as to sift out similar (monoFI) and slightly different (biFI) variants.
I ask you to be sympathetic to this fact.
MetaDriver:
I.e. there is an attempt to "drive" the synthetic into either a "pure trend" or a "pure flat" when compiling the portfolio,
And then have it based on one of the two fairy tale assumptions 1) "If there is a trend, it will continue" 2) "The price always comes back".
My point is that I find the whole thing a bit absurd (well to be honest, just unprofitable). Why? The quotes themselves.
they don't tend to stay in one fairy tale for a long time, they like variety.
What is most interesting is the ability to allow for the probability of market bluffs and almost any kind of crisis in the portfolio. In other words, instead of hoping for a miracle, we should add the situation when the market goes astray to the payment matrix and recalculate it. We get another portfolio, but with a contingency already provided for. And it works strangely enough.
The matter is that when we form the portfolio by history, we obtain it in a form that would have been profitable N bars ago. If we had entered the market then we would have got the full profit without any drawdown. But the train left a long time ago, and if we try to jump in just now, there is a high probability that we will encounter anything: a correction, reversal or intervention of the Central Bank. Therefore, it makes sense to add such a previously unforeseen situation to your portfolio rather than relying on a miracle, according to which the market "will continue" to move in the same vein. The naive hope for miracles of market "stability" tends to fail - financial instruments are non-stationary.
What is most interesting is the ability to foresee the probability of market bluffs and almost any crisis in the portfolio. That is, to put it simply, do not hope for a miracle, but add to the payment matrix used to calculate the portfolio the situation when the market goes astray and recalculate it. We get another portfolio, but with a contingency already provided for. And it works, strangely enough.
If possible, specify how to make such a different portfolio, which will already include the probability of market bluffing.
Elementary, Watson. Flip one of the plots (bars) 180 degrees in the historical data. And recalculate. For example, if we make a payment matrix value[][] for the portfolio, in each cell of which we put, value[i][j] = iClose(s[j], 0, i) - iClose(s[j], 0, i + 1], where i is bar number, j is FI index, s[j] is symbol of j-th FI, then for one of bars - i, all cells are recalculated as: value[i][j] = iClose(s[j], 0, i + 1) - iClose(s[j], 0, i], for all j.
I.e. if the portfolio is generated on N bars, it will be fully consistent with the historical data on N-1 bars and fully contradictory to it on one of the bars. Such a portfolio has much worse performance than the fit, but is no longer a 100% fit.
MetaDriver:
I.e. there is an attempt to "drive" the synthetic into a "purely trend" or "purely flat". Why? The quotes themselvesThe trading strategy does not tend to stay in one fairy tale for a long time, it likes variety.
Some are clearly driving the synthetic into a flat or trend. This, of course, should not be done. Because to drive a synthetic into any curve is a fit.
It's different if you're looking for correlations. For example, you have one FI - EURUSD, and another - also EURUSD. But you don't know their names. You only have their history.
There has to be a method that will find correlations. Such a method has been suggested.
Note, a method for finding correlations has been proposed, not synthetics for trading.
Correlations are defined as weighting factors. For example, if you take the FI with the biggest coefficient and another with the smallest, their cross on history will have the "widest" chart. An elementary example of a spread FI is GBPJPY.
Imagine that GBPJPY is unquoted but there are majors. Who prevents you from finding correlations between the majors and justifiably creating a GBPJPY synthetic based on the principle suggested above?
In the same way, you may create any kind of synthetic on the basis of the found correlations.
You can evaluate the correlations statistically. Look at how coefficients change on the history and draw appropriate conclusions.
Just don't say it's a fit. Otherwise, then everything is a fit, because you always have BackTest in the form of history, and OOS in the form of future.
For some reason there is trust in GBPJPY by many. Why then not similarly trust the history of their synthetic, which could be the same GBPJPY, or something else?
The basis is relationships. And a synthetic (designed to trade) must use them. It should not be flat or trending. It just has to use the existing correlations in a reasonable (statistical) way.
>