This study examines the revealed preference of informed traders to infer the signal quality of earnings announcements. From 2011 to 2015, a cartel of sophisticated traders illegally obtained early access to firm press releases prior to publication and traded over 1000 earnings announcements. I study their constrained profit maximization: which earnings announcements they chose to trade vs. which ones they forwent trading. Consistent with theory, these traders targeted more liquid earnings announcements with larger subsequent stock price movement. Despite earning large profits overall, the informed traders enjoyed only mixed success in identifying the biggest profit opportunities. Controlling for liquidity differences, only 31% of their trades were in the most extreme announcement period return deciles. Using this performance as an empirical moment to a model of informed trade, I estimate that earnings announcements are noisy signals of stock price returns. Specifically, I recover an average signal-to-noise ratio of 0.6. I further explore two potential economic sources of noise: (i) ambiguous market expectations of earnings announcements and (ii) heterogeneous interpretations of earnings information by the marginal investor. Empirically, I document that the informed traders avoided noisier earnings announcements as measured by both sources of noise.