In more than three-quarters of those simulations, the significance test used by GVT reported that there was no reason to reject the null hypothesis—even though the null hypothesis was completely false. The GVT design was underpowered, destined to report the nonexistence of the hot hand even if the hot hand was real.
