MARIO FISHER: Data-driven investing in an era of quick content consumption

Superior data acquisition capabilities and data science skills are becoming key performance drivers for investment managers. As the value of big data grows, traditional funds are using data technology to remove biases from investment decisions, process alternative data and enhance research capabilities and risk management frameworks. Questions remain about how to get the best value out of data and where the risks lie in data-driven decision-making.

Data-driven decisions are made by leveraging verified, analysed data, rather than merely relying on intuition. To extract genuine value from data it must be accurate and relevant to the intended outcome.

If applied correctly there are huge benefits to data-driven decision-making, particularly for asset managers. Data-driven decision-making results in increased transparency and accountability. Data can be stored and studied. This makes it easier to understand why decisions are made and the decision-making process becomes more transparent.

Data also helps with democratising inputs as there is only one source of truth, which leads to better collaboration and repeatable research. Data-driven decision making increases accountability as the data can be accessed before, during and after a decision is made.

Data as an enabler of decision-making can lead to continuous improvement and enhancement of the decision-making process. As the amount of data increases and the technology to consume, process and analyse it becomes more available, the accuracy of the decision improves over time. This type of decision-making does not rely on intuition or gut feel, making it easier to accept and rapidly implement decisions in an era of quick content consumption.

Faced with an ever-growing matrix of information, data-driven decision-making helps with solving complex problems, enabling analysts to test different scenarios and compare outcomes. It also speeds up the decision-making process as the analysis is often done automatically. A systematic investment process, for example, can use real-time data and past data patterns to get valuable analytical insights that significantly increases the accuracy of investment models.

Subconscious biases

Data-driven decision-making creates a natural feedback loop. This means patterns can be identified even before they occur. Using historical data, a process or model can predict what will probably happen in future. This can help quantitative processes navigate ever-changing investment markets and recalibrate as current conditions match events in the past.

It also enables better decisions that are more robust and can be replicated or improved as more data is consumed. Importantly for asset managers, data-driven decision-making helps remedy any conscious or subconscious biases and removes any irrationality in the decision-making process.

The rapid advance in artificial intelligence technology provides potentially promising applications and solutions to data-driven strategies. Data-driven investing is an investment process that analyses traditional and alternative data sources to provide specific investment insights. As more data sets are created, systematic investors can leverage it for analysis, identify patterns, and manage risks to deliver on investment outcomes. New developments in alternative data, such as social media sentiment and news flow, have had the most substantial impact on predictive analytics. Then there is predictive analytics, the practice of applying real-time or current data to help AI models predict risk and return.

Transforming news flow into something usable can be done using natural language processing (NLP). NLP is at the intersection between behavioural finance, linguistics and machine learning. Interest in NLP has risen given the increase in social media and cost-effective computing power used to process it. Sentiment has an influence on stock prices and can be used to determine a predictive relationship with future price movements.

Many approaches to NLP exist, each with positive and negative aspects and varying degrees of effectiveness. Overly complex models seem to be less effective, whereas oversimplification battles in real-world application. Deep learning seems to find the best middle ground, having the ability to analyse and process vast amounts of data required by NLP while still being effective.

Data is a powerful enabler for decision-making, with far-reaching influence regarding driving progress and results. The volumes of data and the velocity of data is increasing at an increasing rate. It is therefore essential for us, as portfolio managers, to harness data and extract meaningful information to guide decision-making in a robust, objective and repeatable way if we are to remain competitive and keep up with the evolving face of asset management as data continues to take us forward.

• Fisher is head of research (customised solutions) at Old Mutual Investment Group.