The hot hand heuristic suggests that, all else held constant, an individual who experienced success in the past is more likely to experience success in the future than an individual who experienced failure. Despite extensive research into the hot hand phenomenon, it is still unclear if belief in such an effect is a "cognitive illusion" or descends from a fast and frugal heuristic that is optimal to use under uncertainty. In this paper, I exploit a rich set of variables in a novel panel dataset to ascertain whether past success affects the probability of future success in a low base rate, high-stakes domain. A rich set of observables allow me to control for non-randomness in success probability. Further, I am able to exploit this structure to test both if individuals tend to exceed expectation over time, and if success increases future expectation. While I find no evidence that above-expected performance tends to persist over time, I find success increases expectation by approximately 14% of the mean. This finding is robust to a number of specifications, and does not fall prey to the novel bias relating to the hot hand identified in Miller and Sanjurjo (2018). I conclude by suggesting the belief that above-expected performance is autocorrelated could driven by the availability heuristic; rare moments of exceptional performance lead to structural breaks in the cumulative residual between observed outcome and expected outcome that can be misinterpreted as persistence in exceptional performance.