One of my main takeaways from judging the @SSFCSpSci data contest is we shouldn't just think of analytics as using one (or both of) event + tracking data. Two of the 4 winners used data sources outside of these two

One of the winners used text-based data and another manually collected set piece signal data to merge with event data. Both provided novel insights without having to be overcomplicated.

There is an almost inevitability of the winning papers at conferences now using applied ML techniques on event or tracking data but there are lots of simple techniques that can be used to get insights from mostly untouched data sources.