Automated trading: A comprehensive guide

Is automated trading a good idea? If you’ve been looking into automating your trading but have been wondering whether or not it is risky or unreliable then read on. In this guide we dive into the ins and outs of this type of trading, and focus specifically on the application of auto trading to stock indices.

We walk through how to set up your trading algorithm, the different trading algorithms for trading indices, and how to design your preferred strategy. So, let’s go.

What is automated trading?

Automated trading – otherwise known as algorithmic or algo trading, auto trading, black-box trading – is the use of computer algorithms to automate the trading process of buying and selling financial assets in various markets. It involves the use of predefined rules and mathematical models to make trading decisions without human intervention. The primary goal of automated systems is to achieve efficient and optimal execution of trading strategies.

How to trade indices

A stock index is essentially a mathematical formula. So, traders often wonder – how is it possible to trade a stock index?

You can do this through Exchange-Traded Funds (ETFs), futures contracts, and Contracts for Difference (CFDs). ETFs and futures contracts are traded on exchanges, while independent market makers such as CFD brokers typically provide CFDs.

All of these instruments enable you to take advantage of the performance of a broad range of market indices, in either direction, without the need to own individual stocks.

Now let’s take a closer look at how auto trading can be applied to trading stock indices.

A brief history of automated trading

Auto trading originated in the 1970s and was further developed in the 1980s when computer technology began to be used in financial markets. However, it wasn't until the 1990s that algo trading gained significant traction. Advancements in computing power and the availability of historical market data for gaining valuable insights enabled traders to develop and backtest complex trading algorithms.

Stock indices, unlike other instruments, have a long historical presence and substantial databanks to go with it. These provide algorithmic novices and experienced traders with plenty of material to backtest their automated strategies with. While some data may not apply to current market conditions, the unwritten rule is that the larger the data sample, the easier it is to build a successful automated trading system.

How to design an automated trading strategy

At this stage, we assume you already know what strategy class you want. So, follow these steps:

Identify entry and exit criteria

Set clear entry and exit rules for your trades and keep these factors in mind:

• Entry signals: determine the conditions or indicators that will trigger your entry into a trade. This may include moving averages, candlestick patterns, or economic events.
• Exit signals: know when to exit a trade, whether based on profit targets, hitting stop loss levels, or activating trailing stop orders.
• Position sizing: figure out the right position size based on your risk tolerance and stop loss level. Make sure you’re not risking more than a predetermined percentage of your trading capital on a single trade.

Backtesting and validation

Backtest your strategy using historical data to see its performance under different market conditions. Keep an eye on profitability, drawdowns, and the risk-reward ratio.

You can backtest using built-in features on the Metatrader 4 or Metatrader 5 trading platform. However, be careful of over-optimization and overfitting. These are common mistakes many traders make when they first start trading. Know more about them in the paragraph below.

What is overfitting?

Overfitting often happens in backtesting when a trader tweaks the parameters of certain indicators or trading rules. The strategy then performs exceptionally well on the training data, but not on new, unseen data, or in real trading situations.

How to avoid overfitting?

Out-of-sample testing and cross-validation.

Trying out your strategy on unseen historical data is also known as out-of-sample testing. Let’s say a trader had developed a strategy based on a 3-year historical period between 2019 and 2022. To ensure the strategy is still relevant, they would test or ‘cross-validate’ that strategy using data from 2023, and see whether its performance is comparable to when it used 2019-2022 data.

The example below shows a machine-based cross-validation test for a trend-following strategy for S&P 500 using Python-based libraries. The strategy continues to generate profit even on unseen data, suggesting it might work well in real market conditions too. The performance of this system in real market conditions, though, is slightly different, though it is still profitable. Our conclusion in this case is that this strategy is not over-optimized for the historical data and has a good chance of working in real market conditions..

An out-of-sample test might reveal that your strategy isn’t suitable for real market conditions, and some ideas may need to be discarded. Therefore, designing a strategy involves trial and error until you find one that works. It’s worth taking the time to do this, as running overfitted strategies in real market conditions isn’t practical.

Running your automated strategy in real time

Transitioning from a historical simulation to live trading involves running a backtested strategy in real time. Here’s how to do it:

Backtesting in a demo or paper trading environment

Most brokers offer a demo or paper trading account. You can use these to test your live trading systems without risking real capital. This helps confirm that your strategy operates as intended under real-time conditions. Consider running it with a small trading account or an Exness standard trading account to ensure proper execution and performance.