Financial service companies have no product to manufacture. For them data is everything. Data is their most important asset. They have millions of transactions each day adding to their data pile.
While ML hasn’t been too successful as yet in finance many attempts are worth mentioning. A major financial institute started analyzing satellite photos to count cars at retail store parking lots so as to forecast their earnings. That is ML at work.
Humans are better at low noise predictions, such as recognizing a human face in a crowd, or reading handwritten zipcodes on an envelope. But they are not so good at high noise situation such as a stock market prediction where S&P 500 has a SD/mean ratio of 20 for example (1).
Use data from overseas stock markets to predict how US stocks would move that day (2). Stock markets, commodity prices, and foreign currency markets are interrelated. Slowdown in the US economy will cause US stock market to drop, but this might make USD or JPY or Gold price to shoot up as people look for safe havens. In paper (2) it is shown that DAX AUD and NASDAQ are correlated and this data can be used to make predictions.