The Role of AI in Driving Improvements in Trade and Investment Analytics
When examining the role of artificial intelligence in trading and investment analytics, it is best to start with what we want to extract from the data. Five of the most common goals include:
1. Identifying what’s unusual in general
We can use AI and machine learning techniques to model the data so that we can readily find anomalies, events or conditions that are unusual and likely interesting. This approach is already gaining adoption because it is relatively mature and it allows traders to manage the deluge of market data and focus their attention on the situations where something interesting and unusual is happening, where human judgement can then add alpha or mitigate risk.
2. Finding what’s interesting to the individual, specifically
AI allows us to describe situations and events that are interesting to us by providing examples, rather than specifying explicit rules. In the past, such opportunities were identified with complex rules in stock screeners, or laborious manual processes to review data. The new techniques make it easier to describe what we want, and they also help automate the process of finding the data that is most useful for the problem we want to solve. Emerging alt-data sets can play a significant role here by augmenting or replacing the factors used in screening.
3. Answering questions using written content
AI, specifically natural language processing, has the potential to radically change how we interact with written content (“unstructured data”), by making it more efficient to search large swathes of text. Imagine a search engine that could answer complex queries about market conditions, margin and volume drivers, that can instantly crunch all available written content and provide an accurate overview of market consensus and the topics that people are focused on and identify the contrarian voices.
4. Automatically surfacing interesting content
Much of the recent progress in AI research in academia and other industries has been around improving personalized content recommendations. In trading and investing, this is a more complex problem, because what’s important and interesting changes rapidly, but the same techniques used to recommend movies and articles can be applied to our domain. This can build on the other three areas to proactively surface relevant and important content based on feedback, so that the system adapts and constantly improves its ability to augment human intelligence.
5. Facilitating Collaboration
For the most difficult questions and uncertain situations, data and analytics are just the starting point, deliberation and discussion among subject matter experts, interspersed with further data analysis, is required to reach good decisions. AI can help by tracking the relationships between points of evidence, hypothesis testing, and speeding up the data analysis cycle.
Tom Doris, Chief Data Scientist