Automated sentiment analysis. I used to teach a segment on this stuff in my data analytics class. Not particularly complex, although you'd want to have a pretty high confidence level, preferably from multiple feeds, before putting a couple of hundred mil on it.
It's called "Machine Readable Data" or "Machine Readable News." It is quite pricey. Reuters quoted me a price so high on the feed that I promptly forgot the cost, on purpose, because I am not agile enough in the mind to make it work for me. https://www.ft.com/content/f9a44a70-0949-11df-ba88-00144feabdc0 And that blurb quoted above is from 10 years ago. Femtosecond microwave tower just millimeters closer to the exchange's receiving antenna than your competitor, anyone?
Dude, you're forgetting about your friends. If there's value in that kind of thing for you, you and I need to talk. Nah, system latencies would swallow up any advantage you got from that. The trick would be having the fastest ETL (extract/transform/load) and, to a lesser degree, fastest processing for multiple data streams. Have your secretary call my secretary; we'll do lunch, and I'll explain everything...
that's interesting. I'd call such market highly correlated market with zero time lag. For such cases, I'd trade multiple futures right away (long NQ, long Dax, long oil, short gold, short US bond, short Germany bond ....).
Well, there's a lot of people spending a lot of money to do exactly what I mentioned in my post... https://www.bloomberg.com/news/features/2019-03-08/the-gazillion-dollar-standoff-over-two-high-frequency-trading-towers And if I could afford a secretary I wouldn't need one, because I would be just wealthy enough to remember when I need to take a break to pee or eat.
Jim Rickards said in a talk that 95% of all trading is done by computers and 80% of all holdings are in passive funds/etfs like SPY.