From Rule-Based Algos to AI-Based Decision Systems
A decade ago, many electronic trading strategies were still mostly rule-based. You could often explain the logic in a few sentences. The systems were automated, but the decision rules were transparent and easy for humans to reason about.
Modern quantitative desks increasingly lean on machine learning and deep learning — and if we want to be a bit buzzwordy, we can call these AI-based trading systems. Models ingest high-dimensional order book data, news, and alternative data, and decisions are made by gradient-boosted trees, deep networks, or ensembles rather than hand-coded heuristics.