More and wider data

Basically, it can be assumed that the higher the level of quantitative trading, the more data needs to be processed behind it. It's the massive amounts of data that are often behind the top quantitative strategies.

Some firms now utilize not only traditional financial data, but also image information such as images of containers in ports captured by satellites, or clues to economic developments from news reports, blogs, and celebrity speeches. With the technical support of image recognition and natural language processing, a lot of unstructured data can also become the object of analysis. Big data, unstructured data and training models all require the intervention of AI technology. patric, head of FRM hedge fund in London, has a good explanation for this: in this age of the Internet, we have access to data that far exceeds the possible processing capacity of human beings. The only way to analyze and recognize patterns in this vast ocean of information is to use machine learning tools and techniques. It's a path to developing a better investment strategy."

Trading strategies that constantly evolve and iterate on themselves

In terms of processing data, AI technology has broadened the sources of data, allowing more data to be included in the analysis. And in terms of algorithms, AI technology also allows financial instruments to automatically evolve and iterate on trading strategies.Alexander, chief investment officer at Rebellion, a pioneer in AI quantitative trading, described his product, saying:

"We gave the system 20 years of global economic and market data, as well as letting it learn about the history of modern finance, and letting it figure out how different factors affect prices across asset classes, sectors and regions. It's not programmed to follow any particular trading strategy because we didn't tell it to look for that. The system automatically recognizes concepts and links them through to performance in specific market conditions."

By contrast, traditional quantitative investing methods tend to strictly apply pre-determined strategies, which are based on the assumption that the present correlation will continue indefinitely. But this tends to cause big problems because markets change rapidly. So the advantage of an AI system is that it is able to evolve its investment strategy as old relationships decay and new ones emerge.

In Rebellion's case, after analyzing financial and trade data, it found that the commodity and FX market cycles have gotten shorter over the last 18 months. So it automatically recalibrates, calculates the impact of shorter cycles, and trades with a new strategy.