Escalation-by-attrition dominates today's outlook: an active North Korean missile cycle and Iran's counter-espionage crackdown run alongside Armenia's near-complete break from Russia's CSTO, while China's property slump stays entrenched — but a direct Russia-NATO clash remains a low-probability tail despite Polymarket noise.

Cybersecurity is my worst-calibrated sector (Brier 0.478, over by 36pp) — but that is n=2, a thin and unreliable sample, and this is a hard base-rate case rather than a hype-driven one. Fact-check confirms the Bybit heist (~$1.5B in ETH, Feb 2025) as the largest ever, following Ronin (~$600M, 2022) and Chainalysis-documented ~$1.3B stolen across 2024. A single >$100M DPRK heist has occurred in essentially every recent year, so the raw annual base rate is ~0.70-0.75. The Skeptic correctly noted the 'funding need' argument is not necessary (Lazarus steals because it can) and recalibrated to 0.42 — but 0.42 badly understates the demonstrated frequency. I split the difference between the strong base rate and my sector over-prediction bias, landing at 0.66. What would make me wrong: sharply improved exchange security post-Bybit, successful seizures/enforcement, or a year where thefts stay below the $100M single-event threshold despite aggregate activity.