Algorithm of multicurrency trading at the market of Forex is realized on the basis of nonlinear stochastic wavelets. The distinctive feature of the algorithm is the possibility of weakly- and strongly connected horizontal self-assemblies, as well as use of nested structures. On-line trading with eight currency couples has shown high effectiveness and stability of the algorithm. It is discussed the problem of possibility of excess profit earning in electronic markets via development of social-financial nets based on synchronization of work of individual traders by means of proposed algorithm.
Certainly, professionals have some reasons to muzz customers of stock strategies. The more complicated system brings more difficult formalization of description space and laws of system evolution. At the same time successful choice of description space allows finding laws, which control system behavior, with minimum costs.
In the present work it is proposed an algorithm of multicurrency strategy optimization for Forex market, however it can be used both for the other markets of shares, options, futures etc.
Nonlinear stochastic wavelet model (NSW model), stated in the works [1,2] has been used as a basis of the algorithm (so called elementary decision units).
Within the framework of this model, the stochastic process
Naturally, the model has non-Markov nature. However, in single-mode case 1 J time independent probability density, find as
, can be represented in the form of quadratures:
, where
Henceforth, the motion equation and corresponding probabilities are restored according to inquiries by means of expansion in a series by suitable set of orthogonal functions (in our case they are Ermit polynoms) using extremum principle.
So, in elementary case, the moment of optimal purchase (entry into the long position) can be
and sales (entry into the short position)
, where 1 is risk level,
are the intent of resale and repurchase moments.
In both cases it is necessary to exclude statistically nonsignificant differences between stationary distributions in time shifts
, and it is fulfilled using Kolmogorov-Smirnov nonparametric criterion.
Henceforth we shall name criteria associated with the sign of 1 dy as elementary dynamic ones and ones with probabilistic assessments as elementary statistical criteria.
The algorithm proposed is suspension over elementary models, which permit as far as it possible to simulate effective decision-making. The control flow chart is given in Figure 1.
Elementary decision generators are specified as 1 and 2, and they conform to statistical and
Total unit number is not limited; and here standard elementary models, such as Moving Average Convergence/Divergence (MACD) method, Bollinger Bands (BB), Market Relative Strength Index etc., can be used.
The following principles are used as a basis of the algorithm:
- Dynamic optimization with the possibility of generation of random solutions, which “survive” or “do not survive” depending on their effectiveness (“creativeness”), unit 4. Here Boolean algebra elements are used, i.e. weighting coefficient of elementary decisions for entry/ exit in to short or long position is taken in such a manner as to minimize norm function of differences between sign of quotation change and decision made as per dynamic and statistical criteria.
The necessity of generation of random solutions nearby local optimums is associated with nonidentity of local efficiency for given currency couple (e.g. maximum profitability of trading within the specified period) to global optimality/ profitability for the package of all currency couples.
- Possibility of tightly-and loosely-coupled horizontal self-assemblies, i.e. automatic connection or disconnection of various program units, directed on various currency couple trading (Unit 6).
The decision, made in Unit 4 for every currency couple, is summarized considering coupled correlation functions. For example, if two currency couples have positive correlation coefficient, then the decision made in Unit 4 regarding entry in to short position as per one currency couple increase probability similar to decision per the other currency couple and vice versa in negative correlation.
At that the units can be in different states that are active trading (decision-making and mutual interaction with trade server -Unit S), semi-active state (discussion of decisions and interchange of information with the other units) and passive state (when the units do not take part in decision making and discussion). Transition between the states is determined for each currency couple as per the quality of prediction of the future quotation change (Unit 7).
Tightly-coupled self-assembly enables to make the collective decisions concerning market entry/ exit on the basis of the current quotations in specified time scale, and in case of looselycoupled self-assembly -on the basis of effectiveness of completed activity in various time scales, Unit 8.
Loosely-coupled self-assembly simulates realization of mixed strategy of work [2], i.e.
compromise between maximum profitability and minimum risk: package reallocation takes place (parts of means invested in short or long positions of various currency couples), Unit 5.
Possibility of self-homothetic assembly enables to use the whole structure completely as the unit of elementary decision making, but in the other time scale (with other time frame), Unit 3.
Possibility of vertical self-assembly: at a time the decision achieved (specifically, entry or exit in short or long pos
This content is AI-processed based on open access ArXiv data.