1. Introduction to Automated Trading Systems

Users set up an automated trading system (ATS) consisting of stock decision-making models and pre-established management rules. Then the ATS will automatically execute transactions according to the parameters users pre-set. We also hope that the research report will overcome the attributes of existing research barriers to become a useful tool for commercial customers who have the need to automatically invest in forex with a small investment size. Furthermore, we hope that this will be a guide for future AIS developers and the foundation for future studies on business models.

Forex is the world’s leading financial trading platform, offering more than 1,000 assets for clients to transact on. The rapid development of the financial market causes an increase in interest in participating in this field. However, more than 60% of transactions are operated not directly by individuals but by banks or law firms holding clients’ money. This report will present automated trading systems (ATSs) that can help investors automatically execute forex orders in the financial trading market. The objective of the research is to design an automated investment solution that can save time for private investors and protect them from psychological pressure. This solution should be implemented on the AIS platform to help investors automatically transact with constant risk management.

1.1. Definition and Purpose

These systems guarantee strict discipline, consistency, and risk management regardless of external circumstances. They use proprietary mathematical models to optimize trade size and timing parameters on a continuous, real-time basis such that the net risk-adjusted return (or the net discount rate per marginal unit of risk held) is maximized. Such trading systems are designed to thrive, regardless of the tricks used by the market to temporarily thwart them, if traded using a modest amount of equity across a broad spectrum of asset classes. They must profitably navigate bull markets, bear markets, and consolidation periods if they are to be considered truly robust. They must demonstrate the ability to generate profits in all market conditions with a minimal level of fluctuations, late delivery, and trade execution inefficiencies.

An automated trading system is a complex program by which a hedge fund or a group of investors manages and executes its basket of trading strategies. The primary focus of the trading system is to provide both superior performance and lower volatility as compared to traditional buy-and-hold investing approaches. The net modified Sharpe ratio, or the risk-adjusted performance of the trading system, must always exceed that of the passive investor. Automated trading systems are used not only for stock trading but also for buying and selling exchange-traded funds, managing private equity investments in early-stage companies, and buying and selling other asset classes such as futures, options, and RTOs.

1.2. Benefits and Drawbacks

Unfortunately, the desire to make money by investing in forex can blind investors to the extent of the losses. The assumption that the use of the program will eliminate simulations and decrease provider service strength over time is another potential problem. Automated trading systems may be a major problem when systems adapt to market conditions, but not always. Algorithm trading systems are less efficient if there is a rapid change in market circumstances and do not respond to this impact. Also, with the number of Forex trading systems, they may believe that traders are hesitant and require guidance, so the program may make instant assessments that can actually result in production decisions. The sheer volume of information that the investor depends on a computer to decide can actually make it harder for them to retain an ostentatious personality and intelligence.

Drawbacks:

One of the most attractive features of automated Forex trading systems is that the “black box” that provides signals can be set up to use past data to effectively predict the likely outcomes. The actual mechanical device can then follow the trends for currency exchange and enter in the forex bid and ask prices, in a nutshell. This market perspective makes a decision without considering the stress, nervousness, or additional stress. Automated trading systems may take the time for business and buy and sell. It is easy to conduct business 24/7, especially when using the best forex method. Forex autopilot programs help beginner investors and those who are more hesitant to find a stroke and make money. Such a computer program allows investors to take part, even if they are in their bank accounts or are sleeping.

2. Key Components of Automated Trading Systems

An automated trading system can be found between the terminals of market participants and trading locations. In contrast to orders placed online, orders that go mechanically via an automated trading system can be entered at a much higher speed than orders placed by a trader using only the monitor and keyboard. As a result, features such as lightning-fast response for many different markets are achieved, which is a significant advantage in the highly competitive market of all electronic exchanges.

A typical electronic trading system (ETS) transmits online orders to buy and sell financial instruments. If the ETS has some level of intelligence and implementation on the client side, then this system is called an automated trading system. An automated trading system is a

computer-based system that sets the parameters of an order, such as the timing, price, or quantity of an order. With the use of such tools as the system, it systematically generates profit potential.

A trading system can be either manual or automated. The manual system requires constant monitoring of the market, while automated systems can manage and implement trading rules without human intervention.

In order to understand the essence of automated trading systems, we need to consider the components of the meaning of the whole term. Let us start with the definition of a trading system. By the term ‘trading system’, we mean a set of conditions for opening and closing positions that are assumed to be profitable. ‘Profitable’ means producing a positive mathematical expectation when repeatedly applied to the full database of historical price series. Many trading systems offer very simple conditions: buy and sell only when a specific price level is reached. At the same time, some simple systems, such as those used to trade carry positions or trend-following counter-trend strategies, are sometimes profitable. However, the most complex systems include some logic for the integration of data from various research areas, such as monetary theory, microstructure, and macroeconomic data.

2.1. Basic Definitions

2.1. Trading Algorithms

The security attributes being used as input for the trading algorithms can be described as direct-income statistical data that can be typically extracted from historical time series data. These data can be related to different types of stock, bond, or derivative prices and can include dividend yields, earnings-to-price relationships, or bond yields. In the case of the Forex market or currency market, input attributes usually include spot and forward exchange rates and exchange rate volatility measures. The effective use of volatility and, hence, the ability to forecast the values of such attributes can provide valuable numeric predictive information and high profits to the trading algorithm. Additionally, the so-called indirect market attributes are forecasting and statistical measures that can also be computed from historical time series data and also be significant on different scales, near or long term.

Trading algorithms are formulated in terms of the annotation of market data and generate deals to buy or sell currencies. The manner in which historical data is applied (or back-tested) to establish the merits of a given trading algorithm is another concern. In recent years, the application of forecasting procedures to financial data has undergone several changes, from the use of single-equation econometric models to the more general and flexible class of techniques known as data-driven approaches. These techniques, mostly classified as what is referred to as artificial intelligence, are designed to allow the data to speak for themselves without strong a priori assumptions about the mechanism generating the data. When these techniques are employed in trading systems, they usually define a decision process that is computed by crossing a threshold or similar rules, commonly selecting the type of annotation that is based on a fractal approach.

2.2. Risk Management Parameters

Spread control with expert advisors is important, and it is difficult to get expert advisors that can manage spread control efficiently. Several top spread controllers were seen in some advisor decompilations, and if the developer is smart, spread control and brokerage conflict-loss collisions settings can be a plus in the EA. However, FapTurbo or Forex Hacked are the classical examples of EAs that control spread properly, but while FapTurbo has professional consulting inputs, Forex Hacked has consulting inputs that are not serious, and one consulting input considerably increases the number of trades by more than twice, making it irresponsible to use them.

Another strong point can be the ability of an EA to manage risk: filters, protections to avoid scalping in a ranging market, protections to avoid overtrading in periods of decreasing volume, dynamic lot management according to account size, money management that uses fixed lots or fixed percentages, inherent protections in an expert advisor, etc.

3. Popular Automated

Trading Strategies

It is important to understand that even within the same class of trading strategies, there are numerous variations. A presentation of popular strategies naturally leads to the topic of this thesis, since the typical frequency of these trades makes human trading very expensive relative to the implementation of machine learning techniques. By fine-tuning the learning algorithms, models with reasonable error levels can be developed even on small datasets. It stands to reason that the use of almost arbitrarily complex models would unearth every existing pattern, even if they happened to be tiny anomalies with only theoretical and no economic significance. By combining signals from multiple indicators, models with higher complexity have better predictive power and robustness, and the essential requirements of trading rules result in very large transaction costs.

The market is dominated by both institutional traders on big banks and hedge funds, as well as individual investors keen on speculating the movements of the world’s biggest market. It is ironic that, given the sophisticated nature of the largest market in the world, a large majority of retail investors rely on simple trading principles, typically hold positions between a few minutes and several hours, and are typically characterized as momentum or trend-following traders. They simply trade in the direction of the trend based on recent historical price movements and expect these movements to continue. Because of the size and visibility of FX flows, momentum trades typically yield significantly better income for traders than they do in other markets. Given that the concept of momentum trading has widespread popularity, it is not shocking that a majority of studies on automated trading strategies and pattern detection focus on strategies that exploit this behavior.

3.1. Trend Following

In a trend-following strategy, the trading signals may occur more frequently in the trending direction than in the opposite direction. There are also gateway positions where the trading system can go short or long depending on the move made by the markets. For example, the values of some of the indicators may go above or below a critical level, triggering a buy or sell order. Such is the case with the Bollinger Bandwidth indicator. This approach, called the dual moving average, is one of the simplest trend-following techniques, usually based on the crossover rules of simultaneous analysis of several averages. Each time the shortest average crosses above or below the longer average, a predetermined quantity leads the system to consider a buy or sell position.

In their basic form, automated systems are an entry tool, specifying the price and conditions under which to enter the market, ideally in the right direction to make the expected profit. They are usually combined with other exit tools, such as stops or profit targets. The most common and simple-to-understand strategy is trend following, where the system follows the market’s movement and opens a position in the direction of the market once a trend has been identified. Statistically, trend-following systems tend to be very profitable due to the observable long-term trends in the market.

3.2. Mean Reversion

Whether a trade is long or short, or neither, for both the buy trade and the sell trade, is determined by taking into account whether the parameter is watched from above its best performance case or from below. Additionally, the parameter is also watched multiple times, less than the maximum allowed, to filter future trades. If both parameters show the correct values, a trade is possible. If a trade is assumed, the position of the trade is determined by checking if the best performance case of the count of parameters watched multiple times has been reached by the buy trade or by the sell trade. If the current watched value is positive, a long trade is assumed, and if the current watched parameter is negative, a short trade is assumed.

For mean reversion, a trade with a negative position from a buy trade and a trade with a positive position from a sell trade are added to the random portfolio. The constraints of unit portfolio positions and the holding period set to the first value of the parameter space not to exceed the available time series outcome are also set for the momentum strategy. Long and short trades for the mean reversion strategy are initiated when prices exceed specific thresholds, which are determined from the available time series outcome.

4. Choosing an Automated Trading System

Before signing any contracts with the automated Forex trading system provider (as, in general, these are service contracts), the trader should inquire about the trading system’s operational terms in terms of market hours of operation and currency pairs traded. The reason for this is that trading systems can only generate profits when they are following a deal, so if the trader is looking for full-time market coverage, he or she needs to confirm with the trading system provider whether the offered trading services effectively cover all desired market hours, adapting to the trading system’s schedule. The currency pair to be traded is fundamental in terms of any agreements made with a trading system provider since most trading systems just deal with some of the most liquid currency pairs.

As with the purchase of an expert advisor, before acquiring an automated forex trading system, it is necessary to collect as much information as possible about the trading system’s performance parameters. The information can be obtained through third-party sources, and if possible, it is also necessary to acquire favorable recommendations. Trading system providers usually offer a test period in real time so that traders can try out their services and see for themselves the potential presented by the trading system under analysis. Although this test can be biased in the sense that the trading results generated may be abnormally high or low in that time period, the test period still allows the trader to learn how to monitor the trading system’s performance. The trader can also benefit from the services a trading system has available at the brokerage firm that provides the trading system. Normally, trading systems are provided through partnerships between brokerage firms and third-party software developers. Those partnerships strengthen the trust that the trader negotiates with the trading system provider and also enhance the service provided for the trader.

4.1. Factors to Consider

There are three ways to consider automated trading systems for trading: independent, semi-independent, and dependent systems. Many retail traders dream of leaving their jobs and making easy money by enabling an expert advisor. Making money is the first struggle. The system has to be robust and able to take advantage of the various market shifts and be able to hedge continuously. It has to have real trades to make money. Programmable ATSs that are based on simple rules are generally not profitable. The key is to make the trading process systematic, and it is possible to realize a profit.

As such, the forex market presents more opportunities for retail traders to (manually) execute automated trading systems (ATS, Expert Advisors) and profit from trading in the forex market with considerable ease and less hassle. These systems largely rely on price patterns and time horizons to generate their trade signals. The trader is relieved from the monotony and pipe dreams of subjective analysis. Automated trading systems make it possible for traders to participate in many markets or multiple strategies at once. The speed of signal transfer, analysis, and trade execution happens in microseconds, and the result or position is typically initiated instantaneously, in contrast to manual trading. In a trading environment, this is a benefit. But there are downsides that will be discussed later.

The forex market is the most liquid financial market in the world. This makes it an attractive trading platform for many traders, although it presents uniquely different challenges compared to other financial markets. The forex market is an over-the-counter (OTC) market. This means that the market exists as an electronic interbank market. Almost every aspect of the forex market is electronic. Information and transactions are available online, almost entirely computer-driven.

4.2. Backtesting and Optimization

For backtesting and optimization, some software tools are very common, like Python,

MetaTrader 5, QuantConnect, and Quantopian; Zipline, which is a Pythonic Algorithmic Trading Library; PyAlgoTrade; QuantLib; AmiBroker; MultiCharts; Zorro; and TradingBlox; and an efficient software tool named NinjaTrader. These platforms have their own characteristics and can adapt to the particular constraints of the developer. For example, MetaTrader and the MQL5 (MetaQuotes Language) were developed primarily for trading forex, and custom indicators are the basis for building trading algorithms. Trading Blox is an open systems platform for efficiently running a wide variety of unattended, fully automated strategies in a Windows environment. Quantopian is, in practice, a blend of Python and the Algo Trading platform. Other tools mentioned were developed primarily for the development of trading strategies for other instruments like equities, futures, and options. The author considers Python and MetaTrader to be the most suitable tools for the development of sophisticated trading models in forex. However, every piece of software has its own characteristics and can adapt to the particular constraints of the author.

A trading system must be optimized in order to find the settings that perform best in a given time frame. This can be done by establishing the set of rules that define the system, the trading signals being generated based on historical data, and the conditions for entering and exiting the market. The trading signals of the trading system are compared to the actual market prices in order to estimate the hypothetical results of trading according to the generated trading signals. A backtesting process can be performed with testing and training samples to identify future profits and risks. Real data can be used and differ from one sample to another. It is necessary to optimize and backtest the trading system and find out which sample to use in a separate validation step for performance estimation. used out-of-sample strategies where both the selection of model parameters and ‘indicators’ were optimized using a separate in-sample period. They found that combining a trend-following and volatility-targeting strategy returned higher median out-of-sample Sharpe (1966) and Sortino (1999) ratios compared to the baseline strategy. Consequently, both optimization and backtesting are possible multiple times, and investors can overfit their trading system, as claimed in the literature.

5. Regulation and Ethics in Automated Trading

Automated trading systems can help regain control over traded volume if the deployment of error handlers, such as automated trading pauses and automated order cancellations, is encouraged. A conflict of interest between the trader and the infrastructure when deploying automatic trading makes further investigation worthwhile. In addition, the implementation of certain restrictions (time and value limits, rate limits) leaves the trader in control and does not impede the trader from automating trading execution. It will also be worthwhile to consider the outcomes of the regulation of automated trading on competition. The paper offers a starting point for further discussion of the role of time on order emergence and its regulatory implications.

The main text you will want to consult to understand the ethics and regulation of algorithmic trading. It defines algorithmic trading systems, provides an overview of the issues involved, and suggests some possible solutions.