Forecast trends. Identify patterns.
Stock markets are famously hard to tackle by ML algorythms alone. However, there is a market niche in building ML supplementary tools for stock traders. These tools allow the trader to make decisions that are more data-driven and science-based.
Some of these tools are in form of forecasts with confidentiality regions, others are identyfying the importance of existing relationships. Some can encompass a wide variety of data sources to generate exogenous features, others can rely only on incoming stock trends. Some can retain past behavior for a long time, while other approaches can tune to more impulsive buy/sell behavior.