Trading Idea Automation
TABLE OF CONTENTS
Introduction
Over the past 20 years, the world of financial investment has been irrevocably changed by the advent of electronic trading.
Indeed, reports from 2018 show that 80 per cent of trades in the US were already being done by machines (1). An increase that has also been experienced in Europe, mostly due to the European Union’s 2018 rule that sought to make deals more transparent by making it harder to conduct over-the-counter trades (2). This, by default, made electronic trading in Europe a great alternative.
This new way of trading, however, is by no means limited to the Western capital markets. In as early as 2010, in a speech at the Paris EUROPALACE, Kiyohiko G. Nishimura, economist and the then-Deputy Governor of the Bank of Japan, told how electronic trading was slowly but surely gaining ground in the country (3).
What all this shows is that electronic trading is not just a passing fad. It’s here to stay; and businesses that work in finances or investment which do not use this avenue are bound to be at a disadvantage. Yet electronic trading has also spurred the rise of trading automation, software that can autonomously trade on your behalf.
This article focuses on the latter, but the first step is to understand what electronic trading is and how it’s changed the game.
Key Points:
- Electronic Trading is the trading of investment instruments over the internet.
- Automated Trading Software can autonomously trade on your behalf, using parameters set by yourself.
- Good Automated Trading Software can download and understand data, make decisions, and even learn as it goes along.
- Off-the-shelf Automated Trading Software often doesn’t let you finetune the way data is being interpreted by the software.
- Companies are now creating customisable, off-the-shelf software that give SMEs the best of both worlds.
What is Electronic Trading?
Electronic trading, which is also known as e-trading, is a simple concept to wrap one’s head around. In essence, it is the digital way of trading investment instruments over the internet. This is usually done through a brokerage account that connects you to an exchange-based system or an electronic communication network.
The rise of electronic trading began in the 1990s, as computers started being connected to the internet. This gave people and companies the possibility to trade stocks, bonds, and exchanges, amongst other things, online.
But, as usually happens with anything related to the internet, the ability to trade electronically has revolutionised the process itself. In fact, trading is now made up of mostly T+1 trades, and dips and surges in the market now happen within intraday time periods.
This means that portfolio managers have to be constantly on the ball; alert to any changes that may affect their investment. Yet technology is fixing the problem it has created by allowing trading to take place through machines, algorithms, and even artificial intelligence (AI) – a process which is called ‘trading automation’.
What is Trading Automation?
As we have seen, electronic trading is the process of trading over the internet rather than over-the-counter. Since these trades happen electronically, however, it stands to reason that there is software out there that can make these processes easier.
Trading Automation is such a type of software, and the easiest way to understand how it works is by looking at another name it goes by, ‘algorithmic trading’. Indeed, trading automation is basically the use of specific algorithms to automatically buy and sell assets, as well as to submit orders to markets or exchanges (4).
This type of software is also called ‘systematic trading’ and ‘robot trading’, but either way it allows for the creation of an ‘Automated Trading System (ATS)’. Such a system allows for huge volumes of data to be processed, as well as larger and more diverse amounts of assets to be traded online at a faster speed. All this can even happen with minimal human intervention required.
This way of trading has now become the norm both for large and small trades. In fact, by 2018, the Goldman Sachs Algorithm was famously handling each and every trade of below $2 million (5). In March 2020, meanwhile, a JPMorgan survey showed that more than 60% of trades for ticket sizes exceeding $10 million were done via algorithmic software (6), too.
All this means that the Algorithmic Trading Market is growing by the day in each region of the world. Indeed, it is estimated that the market of algorithmic trading infrastructure will be worth some $18.8 billion by 2024 (7).
Nevertheless, to those out there who are not well seasoned in the concept of trading automation, it may seem like a crazy idea to let software handle such vast amounts of money. Yet, once you understand how it works, it becomes clearer why it actually makes a lot of sense, both financially and logically.
How Does Trading Automation Work?
We can’t talk about trading automation without referring to algorithms, which are basically lists of steps and instructions something must follow in order for it to reach a conclusion. In theory, these algorithms work in the same way our brain does: in fact, BBC Bitesize has some really sweet examples of how baking a cake or finding the park are two types of algorithms we conduct ourselves in our everyday lives.
Nevertheless, algorithms are usually associated with computers and software: you’ve doubtless heard of how social media platforms, like Facebook, use algorithms to show you the posts you’re most likely to want to see before any others (8).
The question here, however, is how these algorithms are used when it comes to trading automation.
In this scenario, algorithms are used to provide instructions to the trading automation software so that it can decide whether to sell or buy any assets, and when to do so, based on your specific criteria.
Now, that is obviously an over-simplified explanation, but it is essentially how it is done. In fact, the process is automated because in certain trades the computer can take on the role of the broker and make decisions autonomously.
But there are various factors that change what trading automation can do and how much of the process it automated.
Criteria
The most important of these is the criteria, which are pre-determined entry and exit parameters that tell the software when it should do what. Of course, these parameters can – and usually are – different, not just for each portfolio manager, but also for each type of asset being traded. These criteria are set by humans and can range in scope. An example would be parameters which tell the software to sell Stock A when its value plummets below a certain amount (known as ‘stop-loss criteria’), or to buy it when it experiences a certain increase in value.
Decision-making
Then there is the decision-making process, which gives the software the autonomy to make decisions on your behalf based on your selected criteria. This is done following mathematical calculations and automated reasoning, which essentially juxtapose what is happening in the market with your instructions.
Data
Of course, what we need to keep in mind here is that such software requires real-time information in order to make the best possible decisions. This means that trading automation has to work hand-in-hand with data processing, which is an operation that sees the software retrieve data from one or multiple sources, transform it into numbers it can use, and then classify it depending on your parameters. It must be said, in fact, that without this ability, trading automation would not be possible.
Machine Learning
Finally, it’s also good to note that Automated Trading Systems can come with machine learning capabilities, which means that they learn as they go. This AI function helps give them the ability to become better at the process, make more complex decisions, and even predict changes (9).
Benefits of Using Robot Trading
Trading automation comes with a long list of advantages that can truly change the way you trade. Here are some of the biggest ones.
Speeding Up The Process
When you first set up an Automated Trading System, you input the ‘orders’, which the system has to follow. This means that the system will buy, hold on to, or sell assets when those parameters have been met autonomously. This speeds up your response time to changes in the market, and can help you stick to your trading plan.
Reducing Errors
The automation system does not understand trading in the way humans do; to it, all trading is, is a series of boxes that need to be checked in order for a trade to happen or not to happen. That is actually an advantage, as it means that the system will make decisions based solely on your parameters, rather than on emotions.
Moreover, algorithms do not stall and they do not make mistakes. Unlike human beings, such systems can easily keep up with the fast-paced world of electronic trading and execute orders and trades quickly.
Such a system also reduces the chances of trade slippage as the timeframe between the order being placed and it being executed is shortened.
Real-Time Control & Diversification
Another major advantage is that Automated Trading Systems work at lightning-fast speed, with practically all actions taking place within milliseconds. This, in turn, comes with multiple added benefits.
The first of these is that it gives you real-time control over your trading, with information that is accurate and up-to-date driving all the actions. This makes it a powerful tool in a market that works mostly on T+1 trades.
The second is that the system can work on numerous trade accounts and strategies simultaneously, without the need for more time to think and operate. This can give you the ability to trade and spread your risk over various assets and instruments, and even to have a better safety net.
Preservation of Know-How
Trading Automation, it must be said, has also helped solve one of the most debilitating incidents an Investment Firm can go through. Because the System runs on code that reflects the trading strategy and parameters of the company, these are preserved even when a portfolio manager moves on. This establishes continuity of know-how.
More Time for More Important Work
Another important benefit is that robot trading helps free up valuable time that you or your employees would have otherwise spent monitoring the market, effecting trades, and so on. The system does all this for you, but it does not replace humans. Instead, it gives you more time to focus on tasks that require more nuanced decision-making.
Downsides of Using Robot Trading
Yet, of course, like everything else in the world, there are drawbacks to using trading automation. Here’s what you should keep in mind.
Technology Isn’t Infallible
While technology can do a lot more than we can in a shorter span of time, it isn’t perfect. Indeed, there’s a lot that can go wrong. The system, in fact, can be affected by bad internet connections, crashing, and/or your office experiencing a power loss. These are not absolute, however, as data centres, connectivity, and redundant power supplies can all help avoid this.
Human Error
As we’ve discussed before, an Automated Trading System needs to be set up, or rather programmed, before it can start automating your trading processes. This has to be done by humans, and as such, it is also prone to human error that may result in bugs in the software. Such bugs have caused missing or duplicate orders in the past, which is why it’s important that Automated Trading Systems are always monitored by humans to rectify any mistakes or mishaps.
Moreover, sometimes, the software may be over-optimised, meaning that it should theoretically work, but it does not do so when used in a real-life scenario. This often happens when those coding and setting parameters for the software configure these according to past market behaviour. Automated Trading Systems, in fact, should have fluid parameters that can adapt to market behaviour changes as much as possible.
How to Turn Ideas into Code for Trading Automation
Now that we’ve looked at how trading automation works, and what its advantages and disadvantages are, readers may be wondering how this can be applied to specific trading ideas.
The answer is both simple and complex.
It’s simple because the process is essentially to translate the said trading idea into software code. This code would then have a list of actions to undertake based on several rules it must follow.
It’s complex because each trading idea is unique, and one requires both financial and technological expertise in order to make it work as software.
Even so, there are many experts out there who can turn your trading ideas into code, and this is how they do it.
Step 1: Understanding the Psychology Behind the Trading Idea
A big part of automating a trading idea is to understand what the asset or portfolio manager seeks to achieve, and that is likely to diverge from one portfolio manager to another. Indeed, for programmers and financial experts to provide the professional with an automated process that fits their specific needs, they first need to understand the asset manager’s trading strategy.
This, as anyone in the industry will know, can be affected by multiple factors, including the risk appetite, the type of investment instruments available, the amount of assets that are to be invested, and the investment horizon, amongst others. What this gives us, however, is an infinite amount of combinations that need to be catered for by the automated system.
This, in fact, is where the parameters we mentioned earlier come into play.
Step 2: Creating a Data Stream from the Most Pertinent Sources
Whenever a professional makes a decision on an investment, they use their own valuation and gut feeling – two things that are honed and perfected through many years of experience. But the process of making that decision is, in essence, an algorithm in itself: the asset manager looks at the data available, and then chooses whether to buy, sell, or hold on to the assets depending on the crunching of that data.
This shows that data is a crucial component in any investment decision, and it remains so in an Automated Trading System. Indeed, the system actually handles data very similarly to the way a human being would: it collects it, organises it, scores it, and processes it into information that can support making Decision A over Decision B.
For this reason, finding the correct data to feed the System is another major step in ensuring the software functions as it should.
Step 3: Deciding Which System To Go For
As the saying goes, ‘different folks, different strokes’, and it’s apt for trading automation, because there are many types of systems one can choose from.
The Semi-Automated Approach: Some trading and investment ideas, for example, are only based on data rather than action. In this case, the asset manager has the advantage of having real-time information being automatically processed, but some actions, such as the final confirmation, are still taken by the human being.
The Automated Approach: Then there is the fully-automated approach, which is called either Systematic, Algorithm, or Automated Trading. This is when the software collects and models data, and then makes decisions based on the data and the parameters that have been set by the client.
Moreover, there are many different programming languages that can be used to create a system, including C++, C#, Java, Python, and Matlab. Which one to choose depends on the requirements of the portfolio manager and which exchange-based systems and electronic communication networks the Automated Trading System will be connected to.
A final point here is to take the hardware infrastructure into consideration: software needs hardware to operate, and if that isn’t available, there is not much a system can do.
Step 4: Back Testing, Back Testing, and More Back Testing
Some trading ideas seem great in concept but don’t necessarily work in a real-life scenario. This can be down to many factors including the belief that an Automated Trading System will always return a profit (it can’t and it won’t).
The reason for this is that an Automated Trading System does not work in a vacuum. Instead it operates in real-time and is subject to the complexity and volatility of the financial markets.
The best way to understand this is through a short anecdote.
Let’s take the weather forecast as an example. Today, this is done through algorithms that use mathematical calculations to predict what the weather will be. There are many factors that can change the outcome of those mathematical calculations, but one thing that doesn’t change it is how people react: whether people take their umbrella with them to work or not after the weatherman has said that it won’t rain, doesn’t change the possibility of there being rain.
In trading, however, multiple investment models predicting that a security will increase in value will get many people to buy this security. This will result in the price being affected by the way people react, rather than by other external forces. In turn, it will also force a revisiting of the forecasting by the Automated Trading System, which can lead to further changes in the market. This response to the market, for example, is believed to have caused the Dow Jones Industrial Average to cash 800 points in just 10 minutes back in 2018 (10).
This is why, when an Automated Trading System is devised, it needs to have fluid parameters in place, and be back-tested multiple times.
Back-testing is literally checking how the system would have worked in a past financial scenario where we know the final outcome. This gives the programmers a better understanding of how the system is ‘thinking’ ahead of executing orders, and whether the criteria that have been set are aiding or hindering the asset manager achieve the goals of their trading strategy.
Step 5: Monitoring, Maintaining, and Improving the System
Like any other software, Automated Trading Systems need to be monitored, maintained, and improved over time. The same has to be said about the hardware it runs on. Indeed, as hardware technology evolves, new functionalities can be added, old ones can be improved, and the programming language may need to be changed or updated.
Indeed, while such systems are there to help save portfolio managers’ time and to reduce their chances of missing out on opportunities, they cannot function completely on their own.
This is an important step in ensuring long-term success, because unlike what some people believe, such software cannot make money on its own while the asset manager hits a beach in the Caribbean.
Off-the-Shelf Software vs Customisable Software
To conduct a proper slippage analysis, you require specific Trade Cost Analysis (TCA) solutions. In other words, Microsoft Excel just won’t cut it here.
But why is that so?
Off-the-Shelf Software
- One-size-fits-all software that may not provide the automation you need.
- Doesn’t usually allow you to see how data is being processed, which hinders modifications for better results.
- Since it’s off-the-shelf, it’s usually cheaper.
Customisable Software
- Designed around your trading idea, so you can rest assured that it works exactly as you need it to.
- Allows you to check and finetune how data is being processed and how actions are being taken. This allows for modifications to be made and for potentially better results in the future.
- Tends to cost more.
Who Should Choose Trading Automation?
As we come to our conclusion, there is one final thing that needs to be addressed: who Automated Trading Systems can be used by.
The concept of trading automation is broad, and can help any Asset Management Firm in the world turn trading ideas into code. This helps it take back control of its investment processes in a volatile and fluid market, while also giving it a better chance to claim opportunities.
Yet creating, monitoring, and maintaining such a system requires teamwork, with multiple experts on board from a multitude of fields, including investment, risk, compliance, hardware administrators, and software coders all needed to make the system work smoothly.
This, however, means that for Small and Medium Enterprises (SMEs) that don’t have in-house knowledge or expertise on coding and software, customised Automated Trading Systems can seem like a far-fetched dream.
Nevertheless, there is software out there that offers a solid base on which code, customised to the firm’s needs, can be added. This software is a much better option than off-the-shelf ‘solutions’, and it also doesn’t come with a price tag as large as that of built-from-the-ground software, which larger corporations may opt for.
Indeed, as time moves on, more investment firms are realising that investing in an Automated Trading Systems may cost less than they think, and offer a much higher return.
SOURCES:
Wakett is made up of a team of experts from the financial and software fields. Our articles are based on experience and expertise, as well as primary and secondary sources.
🔝(1) Amaro, S. (2018). Sell-offs could be down to machines that control 80% of the US stock market, fund manager says. [online] CNBC. Available at: https://www.cnbc.com/2018/12/05/sell-offs-could-be-down-to-machines-that-control-80percent-of-us-stocks-fund-manager-says.html. [Accessed 22 Dec. 2021]
🔝(2) New EU rules electrify fixed income trading volumes: survey. (2019). Reuters. [online] 17 Jan. Available at: https://www.reuters.com/article/us-eu-markets-mifid-idUSKCN1PB1M3 [Accessed 21 Dec. 2021]
🔝(3) Nishimura, K. (2010) Electronic Trading and Financial Markets: Speech at the Paris EUROPALACE International Financial Forum in Tokyo. [online] Bank of Japan. Available at: https://www.boj.or.jp/en/announcements/press/koen_2010/data/ko1011d.pdf [Accessed 21 Dec. 2021]
🔝(4) Wikipedia. (2020). Automated trading system. [online] Available at: https://en.wikipedia.org/wiki/Automated_trading_system. [Accessed 22 Dec. 2021]
🔝(5) Wigglesworth, R. and Rennison, J. (2018). Bond trading: technology finally disrupts a $50tn market. Financial Times. [online] 9 May. Available at: https://www.ft.com/content/67e48ae4-4fab-11e8-9471-a083af05aea7 [Accessed 29 Dec. 2021]
🔝(6) www.mordorintelligence.com. (n.d.). Algorithmic Trading Market | Growth, Trends, and Forecasts (2020 – 2025). [online] Available at: https://www.mordorintelligence.com/industry-reports/algorithmic-trading-market. [Accessed 29 Dec. 2021]
🔝(7) Algorithmic Trading HPC & AI Reference Guide. (n.d.). [online] Available at: https://www.delltechnologies.com/asset/en-us/products/ready-solutions/industry-market/hpc-ai-algorithmic-trading-guide.pdf [Accessed 29 Dec. 2021]
🔝(8) Barnhart, B. (2018). How to Rise Above Social Media Algorithms. [online] Sprout Social. Available at: https://sproutsocial.com/insights/social-media-algorithms/. [Accessed 21 Dec. 2021]
🔝(9) Solutions, S.F.D., Enterra (2014). Artificial Intelligence: How Algorithms Make Systems Smart. [online] Wired. Available at: https://www.wired.com/insights/2014/09/artificial-intelligence-algorithms-2/. [Accessed 28 Dec. 2021]
🔝(10) Did robot algorithms trigger market plunge? (2018). BBC News. [online] 6 Feb. Available at: https://www.bbc.com/news/business-42959755. [Accessed 28 Dec. 2021]
