Innovation and Data

Alternative data and artificial intelligence in asset management

The Innovation and Data team is composed of Quant Researchers, Data Scientists, Portfolio Managers and Software Engineers. We collaborate with experts throughout the bank to develop methods that can extract valuable insights from the ever-increasing amount of available data and unlock the predictive power of AI.

The industrial world we have created over the past 150 years seems to be changing. In this world, the standardisation of processes played an essential role. Now, however, more and more tasks can be solved with the help of computers or algorithms. With the consequence that human labour and time are no longer limiting factors. After all, algorithms can run in large numbers 24 hours a day. As a result, human efficiency and the share of turnover per employee in traditional companies, such as the automotive industry, are declining in comparison.

The change brought about by artificial intelligence (AI) is happening rapidly and is affecting more and more industries - including asset managers. Berenberg responded to this challenge at an early stage. The Innovation & Data Team was already established in 2018. With this team, Berenberg focuses on machine learning, a sub-area of AI, as well as on alternative data. Alternative data is obtained, for example, by analysing written and spoken news from newspapers, television broadcasts or social media. This data is then made available to the algorithm. This is the basis for one of the pillars of Berenberg's AI-based investment approaches. There, 3.3 billion euros in assets are managed in strategies whose control is based entirely on the machine learning algorithm developed in-house. This evaluates around 600 million documents every month in order to generate daily trading decisions. This is how both the Berenberg Sentiment Fund, a mutual fund with a global macro strategy, and a number of special mandates for pension funds and insurance companies are managed. This application is based on specialised or narrow AI (Artificial Narrow Intelligence), in which an algorithm is trained to solve exactly one question. In this case, the derivation of buy and sell decisions from millions of sentiment data that pass through strict quality filters and pattern recognition.

In parallel, Berenberg has also been working on Artificial General Intelligence (AI). This topic has been on everyone's lips at the latest since the success of ChatGPT and other so-called large language models. In order to create an application possibility, Berenberg has entered into a cooperation with the technology group Google. The aim of the cooperation is to provide the discretionary portfolio managers at Berenberg with a kind of chatbot. Discretionary portfolio managers make investment decisions themselves, based on their own research, company reports and third-party research. But they can't read everything; the AI can gather incomparably more information. The AI can then answer the portfolio manager's questions via a large-language model. The first version of this model is to be launched at the end of 2023, beginning of 2024.

Technical approach

Access to Alternative Data

90% of the world's data was generated in the last two years with 2.5 quintillion bytes data being created each day. Real-time news is captured and interpreted via natural language processing (NLP). These news can be analysed with Machine Learning Technologies, e.g., to extract the underlying market mood or sentiment.

Advanced Machine Learning Techniques

State of the art machine learning models are generated to detect the underlying patterns in large and multi-dimensional data sources. Those patterns are then transformed into investment signals and provided to portfolio management in order to improve discretionary decisions.

Financial markets and technology expertise

The combination of technology as well as financial markets professionals allows for the full leverage of the potential brought by new technologies. Investment processes are tailored for a high functional combination of human experts and AI solutions.

Artificial intelligence is superior to humans in handling and processing very complex and large amounts of data; on the other hand, humans are more flexible and creative. Thus, humans and machines complement each other in a very meaningful way and the combination of both can deliver results that would not be possible on their own.

Nico Baum, Head of Innovation & Data

Investment opportunities

ALOS - Autonomous Learning Overlay Strategies

ALOS is our proprietary AI-based model and strategy generation method to systematically predict future price movements. Building on the interpretation of sentiment data, it has proven to provide strong and consistent performance. The strategy is already live for a multi-year period and various underlying assets. In total 3 billion euros is managed. The Strategies are characterised by low drawdowns and strong diversification effects to classic investment strategies. We would be pleased to advise you in a personal meeting on how ALOS can fit into your personal investment strategy and what opportunities could arise from this.

Berenberg Sentiment Fund

Digitalisation has reached all areas of life and leads to an exponentially growing amount of new data every second. State-of-the-art technology enables us to turn this unstructured data into actionable insights and use these for the prediction of price movements in financial markets: The new Berenberg Sentiment Fund, expands our product range with a market-neutral fund based on our proprietary in-house Big Data and AI approach. Several hundred thousand global and unstructured real-time messages are analysed daily with latest Machine Learning technology and transformed into investment decisions, this process is fully automated. The fund invests in global macro markets and aims to generate attractive and market-neutral returns across the entire market cycle. This characteristic makes the fund suitable as a complementary portfolio component for diversifying traditional asset classes. The sentiment obtained in this way then form the basis of the allocation decision.


Nico Baum
Head of Innovation & Data
Phone +49 69 91 30 90 -462