Tools chosen: kalshi_direct kalshi_related polymarket_related claude_news wikipedia code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a low-probability tail-risk question about a magnitude 8.0+ earthquake in California by end of 2028. The key drivers are historical base rates from USGS data, current seismic activity, and the market's own price. Math on Poisson base rates is essential.

# Event Will there be an M8.0+ earthquake in California before Dec 31, 2028? # Outcomes to forecast - **Yes** (M8.0+ earthquake with epicenter in California or territorial waters before Dec 31, 2028) - **No** --- # Kalshi market anchor **Current YES price: 18.00%** (KXEARTHQUAKECALIFORNIA-28) - 7-day change: +1.00% | 30-day change: +0.00% - Price range (29 days): 10%–18% - Avg daily volume: 61 contracts - **This is substantially above the Poisson/base-rate model estimate of ~2–5%** Related: KXEARTHQUAKECALIFORNIA-35 (before 2035) = 30.00% (-23% over 30 days, high volatility) --- # Sub-question answers 1. **Historical base rate of M8.0+ in California since 1900?** — Zero confirmed M8.0+ events instrumentally recorded. The 1906 San Francisco quake is estimated M7.9. The 1857 Fort Tejon is sometimes listed M7.9–8.0. Historical rate: ~0–2 events in 168 years → annual rate ~0–0.012/yr. [SCEDC, Wikipedia] 2. **Current Kalshi market price?** — 18.00% YES. [Kalshi direct] 3. **Recent significant California seismic activity 2024–2025?** — Dec 5, 2024 M7.0 offshore Cape Mendocino (largest U.S. quake in 2024). California had 7,814 M≥1.5 earthquakes in 2024, more than any year since 1988 per Dr. Lucy Jones. No M8.0+ occurred. [claude_news] 4. **USGS UCERF3 forecast for major California earthquakes?** — UCERF3 gives ~7% probability of M8.0+ in 30 years (from ~2014); implies ~0.24%/year annual rate. For M7.8+, ~60% over 30 years (~2.4%/year). For strict M8.0+ over ~3 years: **~0.7–2.5%** depending on assumptions. [UCERF3/conservation.ca.gov] 5. **Time remaining and Poisson-implied probability?** — ~3 years remain. Poisson estimates: conservative (1/200yr) = 1.49%; moderate (1/120yr) = 2.47%; liberal (1/100yr) = 2.96%; 2-event historical (2/168yr) = 3.51%. UCERF3-implied M7.8+ upper bound = ~8.76%. Central estimate for strict M8.0+: **~2–4%**. [code_execution] --- # Key facts (high-confidence, factual) 1. [SCEDC/Wikipedia] No confirmed M8.0+ earthquake has occurred in California in the instrumental record (post-1900). 2. [UCERF3] 30-year M8.0+ probability ~7% for California; implies ~0.7–0.8%/year. 3. [Wikipedia] 1906 SF quake = M7.9; 1857 Fort Tejon = ~M7.9 — neither firmly exceeds M8.0. 4. [claude_news] Global M8.0+ earthquakes: ~1/year worldwide; zero globally for 3 consecutive years (2022–2024). 5. [code_execution] Poisson central estimate for 3-year window: **~2–4%** for strict M8.0+. 6. [claude_news] Southern San Andreas Fault segment near LA is >150 years overdue for a major rupture. --- # Cross-market signals - **Kalshi KXEARTHQUAKECALIFORNIA-35** (before 2035, ~10-year window): 30.00% — implies ~3%/year over 10 years, which is higher than UCERF3 M8.0+ base rate but consistent with M7.8+ threshold. - **Kalshi KXEARTHQUAKEJAPAN-30** (M8.0+ Japan before 2030): 65% — reflects Japan's much higher seismic rate; useful comparison showing market differentiates geographies. - **Polymarket**: No active California earthquake markets found. --- # Analyst opinions and speculation - Dr. Lucy Jones (seismologist): Southern San Andreas "overdue" for major rupture; 2024 activity elevated. [KTLA/claude_news] - UCERF3 (official USGS): Multi-fault ruptures increase M8.0+ risk vs. prior models; still only ~7% over 30 years. - The 18% Kalshi price is ~5–9x higher than the Poisson/UCERF3 base rate — suggesting market participants may be conflating M7.5+ risk or applying heavy tail risk premiums. --- # Directional lean per outcome - **Yes (M8.0+ by 2028):** Supported by overdue San Andreas southern segment, elevated 2024 seismicity, multi-fault rupture models. Opposed by: zero M8.0+ in 125-year record, strict magnitude threshold, only ~3-year window. - **No:** Strongly supported by historical base rate (~2–4% Poisson), UCERF3 7%/30yr, global rarity of M8.0+ events. The market at 18% appears significantly overpriced vs. models. --- # Gaps / unknowns - Whether 1857/1906 events would qualify under the exact resolution criteria (magnitude estimates have uncertainty). - No updated UCERF4 release to supersede UCERF3 (2014). - Cascadia subduction zone interaction with San Andreas could produce offshore qualifying event. --- # Calibration anchors - **Kalshi current YES price: 18%** (primary anchor — but appears ~5–9x above base rate) - **Poisson central estimate: ~2–4%** for strict M8.0+ over 3 years - **UCERF3 implied 3-year: ~0.7–2.5%** for M8.0+ - **KXEARTHQUAKECALIFORNIA-35 (10-year window): 30%** — implies ~3.5%/3yr if linearly extrapolated, consistent with models - Kalshi 18% is likely a **significant overestimate**; true probability closer to **3–7%** accounting for model uncertainty and tail risk.

1. Both forecasters under-weight the Kalshi anchor relative to the evidence brief's own calibration guidance. The brief explicitly states the "true probability [is] closer to 3–7% accounting for model uncertainty and tail risk," yet both forecasts sit at the very bottom of that range (5% and 6%), effectively treating the anchor as noise rather than a potential signal worth investigating further. 2. Neither forecaster addresses the Cascadia subduction zone interaction noted in the "Gaps/unknowns" section. An offshore Cascadia rupture extending into California's territorial waters could qualify under the resolution criteria, and this represents a genuinely distinct pathway to Yes that is not captured by San Andreas-centric UCERF3 estimates. 3. Both forecasters cite the M7.0 Cape Mendocino quake as modestly elevating concern, but neither addresses the more consequential implication: the brief notes "zero globally for 3 consecutive years (2022–2024)" for M8.0+ events, which is a notable global seismic lull that could argue for either mean-reversion upward or simply confirms the rarity of the threshold. 4. The forecasters do not grapple with the magnitude estimation uncertainty flagged in the brief — the 1857 and 1906 events are estimated at M7.9–8.0 with genuine uncertainty, meaning California's true historical rate of M8.0+ could be 0–2 events, shifting the Poisson upper bound meaningfully and making the "zero in instrumental record" argument less definitive than both forecasters treat it. 5. Both forecasters correctly identify that the 18% Kalshi price may reflect M7.5–7.8 conflation, but neither attempts to use the KXEARTHQUAKECALIFORNIA-35 (30% for 10-year window) cross-market signal to triangulate. If that 30% figure implies ~3.5% over 3 years, both forecasts at 5–6% are actually slightly above that implied rate, which is internally inconsistent with their stated skepticism of the market price.

5% (65% confidence)

Yes: 5% No: 95%

Base-rate models (Poisson ~2-4%, UCERF3 ~0.7-2.5% over 3 years) and zero confirmed M8.0+ in California's instrumental record dominate. The Kalshi 18% anchor likely reflects M7.5+ conflation and tail risk premium given low volume. A modest uplift for overdue San Andreas, Cascadia tail pathway, and magnitude uncertainty on 1857/1906 lands me at ~5%.

Total pipeline time: 105.3s