I started building an SIR-based model in November 2022 to predict how a new COVID variant called XBB.1.5 would impact the trajectory of the pandemic in the US.
Long story short, I overestimated the impact of the variant by a mile. Actual reported deaths (fortunately) peaked at 600 a day (the very lower limit of my prediction of 600-2200 deaths).
Fortunately, my modeling work has come a long way since then. I guess a failed first try is nothing to get discouraged about!