All (not yet) about Strategy Tester, Optimization and Cloud - page 9

[Sergey Golubev]

I want to remind about the following key article: The Fundamentals of Testing in MetaTrader 5
Why is it interesting?
Because it was explained the modes of backtesting: "Every Tick" mode, "1 Minute OHLC" mode and "Open Prices Only" mode.

[Sergey Golubev]

The announcement: New MetaTrader 5 platform build 3440: New trading account report

[Sergey Golubev]

Good article was published: Population optimization algorithms

When optimizing trading systems, the most exciting things are metaheuristic optimization algorithms. They do not require knowledge of the formula of the function being optimized. Population algorithms involve the simultaneous handling of several options for solving the optimization problem and represent an alternative to classical algorithms based on motion trajectories whose search area has only one candidate evolving when solving the problem.

[Sergey Golubev]

Population optimization algorithms: Ant Colony Optimization (ACO)

Belgian researcher Marco Dorigo has created a mathematical model that scientifically describes the process of collective intelligence in an ant colony. He published it in his doctoral dissertation in 1992 and implemented it as an algorithm.

Ant colony optimization (ACO) is a population-based stochastic search method for solving a wide range of combinatorial optimization problems. ACO is based on the concept of stigmergy. In 1959, Pierre-Paul Grasset invented the theory of stigmergy to explain the nest-building behavior of termites. Stigmergy consists of two Greek words: stigma - sign and ergon - action.

[Sergey Golubev]

Population optimization algorithms: Artificial Bee Colony (ABC) 

For many years, the bee search methods were researched exclusively by biologists. However, the interest in applying swarm behavior in the development of new optimization algorithms was growing. In 2005, professor Dervis Karaboga became interested in the research results. He published a scientific article and was the first to describe the model of swarm intelligence mostly inspired by bee dance. The model was called the artificial bee colony.

[Sergey Golubev]

Population optimization algorithms: Grey Wolf Optimizer (GWO)

The gray wolf algorithm is a metaheuristic stochastic swarm intelligence algorithm developed in 2014. Its idea is based on the gray wolf pack hunting model. There are four types of wolves: alpha, beta, delta and omega. Alpha has the most "weight" in decision making and managing the pack. Next come the beta and the delta, which obey the alpha and have power over the rest of the wolves. The omega wolf always obeys the rest of the dominant wolves.

[Sergey Golubev]

Population optimization algorithms: Cuckoo Optimization Algorithm (COA)

The cuckoo is a fascinating bird, not only because of its singing, but also because of its aggressive breeding strategy, which consists in laying eggs into nests of other birds.
Cuckoo search is one of the latest nature-inspired heuristic algorithms developed by Yang and Deb in 2009. It is based on the parasitism of some cuckoo species. This algorithm has been further improved by so-called Levy flights rather than simple isotropic random walk methods.

The Cuckoo Optimization Algorithm (COA) is used for continuous non-linear optimization. COA is inspired by the lifestyle of this bird. The optimization algorithm is based on the features of the species' egg-laying and reproduction. Like other evolutionary approaches, COA starts with an initial population. The basis of the algorithm is an attempt to survive. While competing for survival, some of the birds die. Surviving cuckoos move to better places and begin to breed and lay eggs. Finally, the surviving cuckoos converge in such a way that a cuckoo society with similar fitness values is obtained.
The main advantage of this method is its simplicity: cuckoo search requires only four understandable parameters, so tuning becomes a no-brainer.

[Sergey Golubev]

Population optimization algorithms: Fish School Search (FSS)

An aggregation of fish is the general term for any collection of fish that have gathered together in some locality. Fish aggregations can be structured or unstructured.
Fish form schools in nature in several ways. As a rule, they prefer larger schools, consisting of individuals only of their own species.
The question of how schooling fish choose the direction in which to swim remains unresolved. This schooling behavior prompted many researchers to create not only a mathematical model, but also an algorithmic model for solving various optimization problems.

[Sergey Golubev]

Population optimization algorithms: Firefly Algorithm (FA)

Nature has always been an inspiration for many metaheuristic algorithms. It managed to find solutions to problems without prompting, based on individual experience. Natural selection and survival of the fittest were the main motivation for the creation of early metaheuristic algorithms. In nature, animals communicate with each other in many ways. Fireflies use their ability to blink to communicate. There are about 2000 species of fireflies with their own special flash patterns.

There are two variants of population optimization algorithms inspired by the behavior of fireflies: the Firefly Algorithm and the Glowworm Swarm Optimization (GSO) algorithm. The main difference between firefly and glowworms is that the latter are wingless. In this article, we will consider the first type of the optimization algorithm.

[Sergey Golubev]

PR calculation for agents

Forum on trading, automated trading systems and testing trading strategies

Rating of PR agents

MetaQuotes , 2019.02.27 09:09

MQL5 Cloud Network statistics: https://cloud.mql5.com/ru/stats

The PR rating is fully correlated with the processor power, since we have calculation tasks:

• https://www.cpubenchmark.net/
• https://www.cpubenchmark.net/high_end_cpus.html
  

and -

Forum on trading, automated trading systems and testing trading strategies

cores disappeared after updating the tester.

Renat Fatkhullin, 2020.07.30 18:43

We upgrade the cloud network every day, changing ratings and adjusting to experts.

Now, instead of pure PR, a complex formula is used that takes into account:

• PR - CPU performance
• RAM size
• free disk space
  
• disk read speed
• ping to the nearest cloud server
• percentage of network losses

This made it possible to drastically reduce the latency from the fact that some tasks fell to openly inhibitory agents that delayed all calculations.

Three cloud servers in the USA, Germany and Russia are currently distributing tasks, which has significantly reduced network latency and accelerated data delivery. Some experts require replication of hundreds of gigabytes, and more than a terabyte for those who excel.

and -

MetaTrader 5 Help → MQL5 Cloud Network → Price Calculation

-----------------

More related:

MetaTrader 5 Help → MQL5 Cloud Network → How to Participate - Restrictions of Participation on MQL5 Cloud Network