Since the beginning of modern economy, financial news have influenced investment decisions. New technologies emerge and investors take this to the next level.
For centuries, stock traders and other market participants have been trying to gain an information advantage through market news. The idea of this strategy: whoever is the first to have explosive information can earn a lot of money with it.
One who recognized the economic potential of news early on was Paul Julius Reuter. Even back then, in 1850, financial news and stock market prices were transported via telegraph lines. However, the telegraph network was full of holes, and the route between Paris and Berlin, for example, had to be bridged in part by mail train.
Reuter saw an opportunity and tried to speed up the news flow. He began to use pigeons, which transported the stock exchange data three times faster. The success of his new service for financial information was enormous and laid the foundation for the news agency "Reuters" - today a global player in its field.
Carrier pigeons have long since been replaced by digital means of communication, but the demand for timely and reliable financial information remains unbroken. Even today, news serves as a kind of raw material for trading strategies. Investors need to know about the latest interest rate decisions by the Federal Reserve, but also about the latest tweets of Elon Musk. Social networks are playing an increasingly important role and influence the global financial markets. Their relevance has increased and they cannot be ignored by investor community. But how can this most diverse information be taken up profitably?
One possibility is offered by Natural language processing (NLP). This technology uses computers to analyze a wide variety of news sources very efficiently. It combines computational linguistics, machine learning and artificial intelligence with the aim of systematically measuring market sentiment through Twitter or RSS feeds, newspaper articles and press releases.
For analyzing the news sentiment, thousands of items from a wide variety of news sources are processed every minute and a sentiment score is determined. Good news receives positive scores, bad news negative ones. The process is complex and also takes into account the interaction of current and previous news. The goal is always to define sentiment values that are as accurate as possible for a wide variety of financial markets. With the help of algorithms, these are converted into return forecasts and trading signals are generated.
One advantage of NLP is the enormously fast and comprehensive, algorithm-based message processing for founding investment decisions. The idea is that the computer "reads" a document with the help of analytics algorithms and then assigns a sentiment value. Here are two exemplary news headlines to describe the strategy for analyzing:
The computer recognizes that both sentences refer to the currency of Great Britain, the pound sterling. It is associated with the words "continues rise", "optimism" or even "climbs" - all of which imply a positive sentiment. As a result of this sentiment analysis, the algorithm derives a positive return forecast for the pound sterling and generates a buy signal for trading.
Andreas Vetsch is a research analyst at LGT Capital Partners and in this role, also monitors the monetary policy of central banks.
He and his colleagues in Asset Management look for attractive investment opportunities and identify the best portfolio managers. They also manage a substantial portion of the assets of LGT’s owner, the Princely House of Liechtenstein.
If you are interested in the latest global market and economic developments, we recommend you read the insights provided by our research experts.
However, the news has simultaneous effects on other currencies, stocks, interest rates or inflation. For example, the two rates lead to a negative signal for British stocks, as a rising currency has a negative impact on exports by British companies. In this case, the sentiment value for export-oriented companies is particularly negative. The algorithm processes these messages and correlations much faster than a human brain and generates an information advantage.
Working with language, however, is more difficult than working with well-structured numerical data. A number can be assigned a unique value, but words cannot. The following sentence illustrates this:
Again, the word "sterling" is mentioned and the word is also associated with "rises". In contrast to the previous example, however, we are not talking about the pound sterling or other financial news here, but about Raheem Sterling, the English national soccer player.
To prevent the algorithm from automatically triggering a buy signal for the pound every time Raheem Sterling scores a goal or receives positive tweets from English soccer fans, NLP must also recognize the news context. Semantics are used to establish references that take place in specific environments.
If "Sterling" refers to words like "Soccer", "Manchester City" or as above to its coach "Guardiola", the algorithm recognizes a soccer environment and distinguishes to the currency environment. This simple example provides an insight into the complexity of systematic message processing.
Used correctly, NLP offers enormous potential. As the volume of news and its processing have changed dramatically since the days of Paul Julius Reuter, today's investors respond to financial news just as they did back then. The same news patterns still have a strong impact on the opinion of the investors.
Fortunately, however, access to information has become much easier. Pigeons have long since been replaced by automated data feeds. That's a good thing - just imagine the enormous amount of pigeons to handle more than 500 million tweets per day.
While the world has changed a lot since then however, investors may still profit from an accurate and fast news-based trading strategy. It can still deliver an information edge today, to achieve attractive market returns. This impressively demonstrates that the spirit of Paul Julius Reuter remains vital, who once said: "First do it right, then do it first".
LGT Capital Partners launched the LGT AI News-Based Trading (NBT) strategy on 1. April 2016. This systematic investment strategy forecasts market movements with financial news as its only input and trades futures on the major equity indices based on these forecasts.
The strategy applies a natural language processing system to screen and score tens of thousands news items per day from hundreds of news outlets and a nonlinear dynamic model to calculate investor sentiment from the scored news. The daily target long/short exposure is then determined using machine learning algorithms.
NBT delivers regime-independent absolute performance with low correlations to hedge fund and long-only strategies. Since inception, it has generated a top-of-the-range track record. In particular, NBT demonstrated an excellent performance in the turbulent 2020 and won the HFM European Quant Award in the CTA Specialist Category.
The news-based trading strategy is part of the hedge fund offering of LGT Capital Partners. Learn more about these investment solutions.