Clock-change crashes (+3.7%) and football-loss market dips (-0.11pp) both unresolvable in re-tests
beglen · x · 2026-10-27
The final two studies in beglen's re-test series (full one-page reports in images): the spring clock change's famous +5.8% fatal-crash spike (Smith 2016) re-estimated at +3.7% with uncertainty spanning zero — "can't tell yet" (NHTSA FARS 2002–2024, 52,128 crashes); and the football-losses stock-market effect (Edmans et al. 2007, -0.26 risk-adjusted points) re-estimated at just -0.11pp across 37 countries, 2006–2024, indistinguishable from zero. Both analysis plans were timestamped (GitHub commit 552293f) before data was pulled.
More from Research
- OpenAI model claims refutation of ETH; Impagliazzo to help vet the proof — StefanoGogioso · 2026-10-27
- Noah AI's free tool turns a research topic into a cited, publication-ready literature review — socialwithaayan · 2026-10-27
- Complexity theory is a dark jungle; simple conjecture formulae may be illusion — StefanoGogioso · 2026-10-27
- Epoch AI audits 15 benchmarks: 4 verified, 9 flawed, and pushback dismissed — hugobowne · 2026-10-27
- Developer uses Claude Code and Codex to find a possible planet — CodeByPoonam · 2026-10-27
- Sabine Hossenfelder: AI will mine hidden scientific discoveries before it replaces truck drivers — skdh · 2026-10-27