In this section, I reflect on forecasts where the outcome diverged significantly from my expectations - whether due to a poor Brier score, an unsatisfactory ranking, or simply missing the positive resolution. The goal is not just to acknowledge these misses, but to dissect what went wrong in my initial reasoning or methodology. By doing so, I aim to identify patterns, biases, or blind spots and use these insights to improve my future forecasting accuracy and calibration.

What will be China’s estimated annual GDP growth rate for 2025, according to the IMF? (31/03/25)

Not exactly a postmortem, yet. At the time of writing (31/03/25), the question is still active with ~17–18 hours to go. But I had a last minute realization that’s worth documenting.

I was just thinking about this question and it hit me - why not look for studies or reports evaluating the track record of IMF forecasts? A quick search turned up a couple of promising papers. (Linked here: https://www.gjopen.com/comments/1972001)

Note to future self: Think. Then think deeper. Or as my favorite book puts it: Think like a Freak.

Will new federal elections be called in Canada before 5 April 2025? (resolved 25/03/25)

Looking back, my initial intuition of 60% (Yes) / 40% (No) was quite off compared to the early forecasts of most participants. While I did adjust frequently, raising my estimate to 85% within three days, many others had already locked in high confidence forecasts much earlier.

The root cause of my initial misjudgment was a lack of understanding of Canadian political dynamics. I particularly underestimated how Trump era tariffs had shaped public sentiment and policy alignment in Canada.

Not to future self: Read up on Canadian politics (this is of course question specific) and be mindful of crowd forecast neglect next time.

Will Travis Kelce publicly announce his retirement from the NFL before 16 March 2025? (resolved 16/03/25)

I was overly cautious with my initial forecast, giving only a 51% chance that Travis would retire before March 16th. My focus was mainly on reports that he was exploring opportunities in Hollywood, which I over-weighted relative to other signals. As a result, I consistently remained below the crowd forecast.

That said, I did update my forecast quickly once the news broke that Travis would not be retiring, which helped limit the damage. In hindsight, I should have placed greater emphasis on the lack of solid confirmation from credible sources and been more attuned to the crowd's early conviction, which was likely grounded in better contextual knowledge.