Algorithmic trading is quite commonly utilized for the buying and selling of currencies in the foreign exchange (forex) market.

Such technology oriented approach is ideal for forex trading given its speed and flexibility. Most importantly, it eliminates the emotional ups and downs of manual trading, which enforces discipline in risk management. 

It is flexible because a wide range of strategies can be deployed through the algorithm.  There are four basic types of strategies that underlie at least 12 forex algorithmic trading strategies. Strategies can be mixed and matched in any number of combinations, depending on the trader’s objective.

In this article, we briefly explain what each strategy entails, starting with the four basic types of strategies. A caveat must be included here that the following descriptions are purely for educational purposes, and by no means ensure profitability. As always, please do your own research and due diligence. 

STATISTICAL 

This strategy relies on an analysis of historical time series data, which in this context, the data here would refer to the bid-offer (buy-sell) price of a foreign currency.

These data points, often being the closing / highest / or lowest bid-offer price, are recorded at regular intervals over a period of time. By analysing the movement of selected data points, one can identify trends and opportunities based on this data.

These historical trends can then be used to compare with the current market data and trends so that profitable trading opportunities can be identified.

AUTO-HEDGING 

As the name suggests, it is a strategy which employs hedging automatically. This means that once a trade is placed, another trade is executed to hedge the first trade.

The main aim is to reduce a portfolio’s risk exposure. Consequently, it could help to balance a portfolio’s exposure across various currency pairs. 

FOREX SCALPING 

Forex scalping involves moving in and out of trading positions very quickly throughout the day. The result is tiny profits from very small market movements at any given time. Thus, giving rise to the term ‘scalping’.

Using algorithmic trading, it is possible to make thousands of trades per day, which is at a much faster rate than manual trading.

HIGH FREQUENCY TRADING

One of the subcategories of algo trading is high frequency trading (HFT). HFT involves the use of algos to make up to hundreds of thousands of trades in a fraction of a second.

Given algorithms are automated, they are able to operate at a frequency and pace that no human trader can. 

Manual traders will be subjected to the limitations of their physical and mental capabilities,  whereas an algo can monitor price movements and operate 24-7 without any down time.

Consequently, algorithms can identify and capture the best prices in milliseconds such as trading on pairs with the highest bid-ask spreads at any time of the day. This maximises the potential profit for traders with greater ease.

Scalping is often used in unison with HFT in forex trading. These strategies enable high volumes of trade on quick price fluctuations. 

DIRECT MARKET ACCESS 

This forex algorithmic trading strategy entails the use of direct market access (DMA) to place trades. DMA means access to a sophisticated technology infrastructure which is often owned by sell-side firms.

This infrastructure connects to multiple trading platforms and houses order books of financial market exchanges.

The DMA strategy is usually used by buy-side firms. Instead of relying on market-making firms and broker-dealers for trading, buy-side firms use DMA to carry out their trades. 

TREND-FOLLOWING / PRICE ACTION

Being one of the most straightforward strategies, this forex algo trading strategy involves following market trends. When a trend is bullish, the algorithm will take a long position. When it is bearish, the algorithm will likely go short.

In order to determine a trend more conclusively, the algorithm compares historical data with current data as a basis to build projections. From there, it can show whether a trend is likely to continue or reverse.

TREND-FOLLOWING MOMENTUM

Aside from just studying the movement of asset prices, there is a strategy which studies the momentum of a trend. It employs the use of technical trading indicators to give signals of when a trend will change. Examples of commonly used indicators are Moving Averages and Stochastic Oscillator.

Moving Averages help to identify a long-term trend. Instead of having to plot the price changes manually, the algorithm can be programmed to track price movements in the form of 20-day or 50-day averages. The shorter the day average, the closer it will be to the most recent asset price.

Stochastic Oscillators are often used as signs of overbought or oversold conditions. They can signal when an uptrend or downtrend is about to occur due to the asset being overbought or oversold.

MEAN REVERSION SYSTEM / AVERAGE PRICE

It is an algorithmic forex strategy based on the assumption that historical returns and currency exchange rates will eventually return to their average levels at some point in the future. With this in mind, the strategy tries to capitalise on the dramatic changes of currency value.

However, it is important to note that there are no guarantees in forex trading and no guarantees of a return to a normal pattern. Events, locally and globally, can potentially throw markets off tangent in the long-term.

The commonly used technical trading indicators used in a mean reversion system are Moving Averages and Bollinger bands. Moving Averages provide the historical average price of an asset while Bollinger bands help to identify a market which has moved too far from an average.

This trading strategy may be more suitable for traders who prefer short trading timeframes, such as one-hour charts, four-hour charts or daily charts. 

NEWS-BASED

Events within a country or in another country can have an impact on the world’s currencies. For instance, political unrest, pandemics, elections, inflation, war and so on.

A news-based trading strategy is programmed to react to news reports. The system is designed to track news wires then generate trade signals based on these events in real-time.

However, events that reach major news outlets tend to be stale as the advantage comes from having information or knowledge to act before competitors. 

Acknowledging this, there are also many technical traders who choose to ignore news to reduce noise in their trade decisions, and choose to simply react to the price action and behaviour of the market for their trades.

MARKET SENTIMENT

This strategy involves using either the Commitments of Traders (COT) report or social media scanning to gather news on market sentiments. It is thus considered a news-based algo trading system.

COT systematic strategy detects extreme net short or long positions. The other information gathering approach involves scanning social media networks such as Twitter to get an idea of currency biases.

This news reveals the actions of other traders and helps to predict future price movement. Manual traders who want to employ this strategy need to have a firm understanding of how the financial markets operate and strong skills to develop sentiment trading algorithms.

ARBITRAGE

This method of trading depends on exploiting price anomalies across different financial markets. Arbitrage used to be more profitable in the past.

Now that technology has become a lot more advanced and sophisticated, price anomalies do not stay for long. This is especially so for currencies, as price differences in forex are usually very small.

As such, the arbitrage strategy has to be carried out by trading in very large volumes in order to make a substantial enough profit.

Under this classification, triangular arbitrage is a popular strategy. It involves two currency pairs and a currency cross between the two.

STATISTICAL ARBITRAGE

Also referred to as ‘stats arb’ strategy, it is a subset of mean reversion strategy.

Most algorithms for this strategy are designed to exploit statistical mispricing or price inefficiencies of one or more assets. That is why the strategy involves complex quantitative models and requires substantial computational power.

The most popular form of this strategy is pairs trading. It is used to trade the differentials between two markets or assets. A long position is taken in one asset; at the same time, an equal-sized short position is taken in another asset.

ICEBERGING

Usually employed by large financial institutions, the strategy involves breaking one very large forex trade into many smaller positions.

The algorithm for this strategy executes trades under different brokers at different times so as to mask the actual volume from other market participants.

Retail traders who are trying to keep track of a financial institution’s trading volume are thus only able to see ‘the tip of the iceberg’. This strategy also enables the financial institution to trade under normal market conditions and avoid sudden price fluctuations.

STEALTH

Because of the common occurrence of iceberging in the past few years, hardcore market watchers have created a strategy to overcome obstacles posed by financial institutions.

An algorithm for stealth trading strategy is able to covertly piece all the small trades by financial institutions and reveal the actual market player behind the trades.

Once the market player has been revealed by the algorithm, traders using this platform can respond in a manner favourable to themselves. 

MARKET-MAKING

The algorithm for the market-making strategy aims to supply the market with buy and sell price quotes. In other words, instead of responding to market trends, it facilitates the creation of a market by quoting prices. 

Aside from that, it is also used for matching buy-sell orders and capturing spreads.

MACHINE LEARNING AI

A relatively new form of algo trading, it utilises machine learning and artificial intelligence (AI). The benefit is that by parsing in huge quantities of data, the algorithm will be able to pick out less obvious attributes that may be profitable. 

In some cases, the outcome of these attributes found can be fed back into the algorithm, allowing it to ‘remember’ and ‘learn’ from its trades done. This will then enable the algorithm to update itself on what has and has not been working.