Pair trading and multicurrency arbitrage. The showdown. - page 61
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
correlations of what with what?
One real character to another.
Not really. The task is to build a synthetic symbol, which can trade a flat TS in plus for a short time after the construction. And it is not necessary to be able to build such a symbol at any moment.
It is enough, for example, to be able to build it in the evening. In particular, evening cross scalping is the ability to build such symbols. Where in fact symbols are already built by quotation machines. I.e. you don't need to build EURUSD^(+0.5)*GBPUSD^(-0.5) yourself, but you can trade EURGBP right away.
With this formulation of the problem, classical triangles are a very special case. There you build such a symbol, which has the following events (not simultaneously): Bid>1 (Sell_Signal), Ask<1 (Buy_Signal). I.e. a very primitive flat TS.
You are talking about another type of arbitrage - IMHO, you are talking about statistical arbitrage. I'm talking about arbitrage in the form of a chain of exchanges, which is impossible in Forex in its pure form, but only as a lock of three pairs, which must then be sorted out somehow (I suggested the simplest way of sorting out, which makes this type of arbitrage similar to statistical arbitrage). Practically, such arbitrage is hardly possible on forex, but they try to apply it on crypto. Above I posted a link to an article in the hub about an algorithm for detecting such arbitrage chains.
It is probably possible to generalise (and even combine) these two types of arbitrage into one.
It is probably quite possible to generalise (and even combine) these two types of arbitration into one.
This is how it is done. Classical is a primitive case of statistical.
Statistical in a general form implies the presence of a flat TS. And sometimes (some indicator is in the range. For example, time of day) the ability to build a symbol to plus this TS.
Classic: a flat TS: below one - buy, above one - sell. The construction is round-the-clock with unchanged weighting coefficients. I.e. the most primitive statistical.
The task is to find a stable cloud. As a rule, such stability is achieved in the evening time, that is why it is scalped for many years.
in the order of thinking and brainstorming:
the cloud is a rather nice, moderately well-fed ellipse. The current count is swooping around it, you might say rotating.
For any window, the centre of the ellipse (motion/rotation) never coincides with the origin. This is a generalised trend, which also flickers when the window shifts, but with a smaller relative magnitude.
Like a matryoshka doll, or like a vortex if you imagine the third dimension.
Since they are all nested together with explicit degrees, you can take ln or sqrt again to remove the nonlinearity :-)
the ellipse will remain an ellipse, just more rounded.
but how to interpret the observations in the "rounded ellipse" :-)
by way of reflection and brainstorming:
The cloud is a rather nice, moderately well-fed ellipse. The current count is moving around it, you could say rotating.
Probably not understanding each other. But I will continue a little.
As I wrote above, it is required to achieve the plus of a flat TS.
As a particular solution to such a problem, we assume that we need a high KK between two rows a[i] and b[i]. Then at high KK the new row c[i] (= a[i]/b[i]) will be a kind of flat oscillation around some level. The higher the KK, the stronger the fluctuation, but the lower the amplitude. In trading language it means that there are a lot of small trades. That is why one tries to take a high CK, but not too high.
But this is just a mathematical partial solution, which is used because of the ability to calculate the QC relatively quickly. In fact, no QC has nothing to do with it. A flat TS can make money even on the quotes in the form of a trend. The QC approach is too strong narrowing of the set of solutions.
mytarmailS #:
Почему КК, а не коинтеграция?
Because CC is cheap (so you can go through a huge number of variants), cointegration is expensive crap with wildly low statistical significance for trade. Cointegration was profitable only when the Nobel Prize was paid.
Stat. arbitrage is not about stationarity. I defined it above.
is achieved in the evening, which is why it's been scalped for years.
Times are not what they used to be. Kitchens are raising evening spreads, and soon "non-cooks" will be forced to raise them as well.