Analytics was always overrated, for one thing, most people, including a lot of people who put out sports metrics, don't understand the limits of statistical modeling. You should use analytics to ask better questions, not to provide definitive answers. And remember game theory, once you make a change, your competitors will respond to that change, and its effectiveness may dissipate. So you go launch angle, pitching coaches stress 4 seamers at the top of zone (harder for launch angle hitters to get around on).
The stock market is similar, they used to find all these anomalies, after a while, half were shown to be statistical flukes, the other half disappeared once they were published and investors arbitraged them away.
When you use analytics to ask the right questions, you start to recognize WHY certain things work, and for which players, and how opponents will counter them, and whether you have a response to that counter. Simply finding correlations isn't good enough.