Tools chosen: kalshi_direct kalshi_related polymarket_related wikipedia claude_news code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a long-horizon seismic hazard question requiring base-rate estimation rather than current news. The key is the historical frequency of M8.0+ earthquakes in California and the geologic plausibility, since California faults (San Andreas) have debated maximum magnitudes. The Kalshi price is the primary anchor.

# Event Will there be an M8.0+ magnitude earthquake in California before Dec 31, 2035? # Outcomes to forecast - **Yes** (M8.0+ epicenter in California or territorial waters before 2035) - **No** # Kalshi market anchor **KXEARTHQUAKECALIFORNIA-35: 26.00% YES** (current) - 7-day change: +1.00% | 30-day change: **-27.00%** (significant recent correction downward) - Price range over 54 days: 22%–65% (high volatility, thin market) - Avg daily volume: 44 contracts — **low liquidity, prices likely unreliable** - Related: KXEARTHQUAKECALIFORNIA-28 (before 2028) = 16% YES # Sub-question answers 1. **Current Kalshi price** — 26.00% YES for before 2035; recently crashed from ~65% high, suggesting initial mispricing corrected. [Kalshi direct] 2. **Historical M8.0+ in California** — Zero confirmed M8.0+ earthquakes in ~166 years of records. Largest historical events: 1857 Fort Tejon (~M7.9) and 1906 San Francisco (~M7.9). [Wikipedia; code_execution] 3. **USGS/UCERF3 10-year probability** — UCERF3 gives ~7% over 30 years for M8.0+; pro-rata ~10-year probability ≈ **~2–2.5%**. UCERF3-derived annual rate for M8.0+: ~0.00147/yr → P(≥1 in 10 yrs) ≈ 1.46%. [claude_news/UCERF3] 4. **Maximum credible magnitude** — San Andreas can produce up to M8.1–8.3; southern segment capable of ~M8.1. Offshore Cascadia (not strictly California epicenter) could exceed M9. [claude_news/earthquakecountry.org, Wikipedia/San Andreas] 5. **Annual base rate** — Global M8.0+ rate ~1–2/yr; California-specific historical MLE = 0/166 yrs. Poisson 95th-percentile upper bound on California rate: 0.018/yr. UCERF3 central estimate: ~0.00147–0.002/yr. [code_execution] # Key facts (high-confidence, factual) 1. [UCERF3/CGS] 7% probability of M8.0+ in California over 30 years (increased from 4.7% under UCERF2) 2. [UCERF3] Expected recurrence of M8.0+ statewide: ~494 years (down from 617 yrs under UCERF2) 3. [code_execution/Wikipedia] Zero M8.0+ earthquakes recorded in California in 166 years of modern records 4. [claude_news] Southern San Andreas last ruptured in 1857 (~168 years ago); recurrence interval ~140–160 years — technically overdue 5. [code_execution] Poisson model: at rate 0.002/yr → P(≥1 in 10 yrs) = 1.98%; at 0.005/yr → 4.88% 6. [code_execution] Weighted central scenario estimate: ~1.97% probability # Cross-market signals - **Kalshi KXEARTHQUAKECALIFORNIA-28 (before 2028):** 16% YES — implies ~10% incremental probability for 2028–2035 window at current Kalshi pricing (internally somewhat inconsistent given science) - **Kalshi KXEARTHQUAKEJAPAN-30:** 58% YES — Japan far more seismically active at M8+ scale; useful contrast - **Polymarket:** No active markets found - **Sportsbook:** N/A # Analyst opinions and speculation - Scientific consensus (UCERF3): ~2–2.5% over 10 years; market at 26% is **~10–13x higher** than scientific base rate - The 30-day price crash (-27pp) suggests the market corrected from a likely mispriced initial listing - Multi-fault rupture scenarios (UCERF3 innovation) slightly elevate probability vs. older models but remain insufficient to justify high double-digit probabilities - Southern San Andreas "overdue" narrative is real but doesn't change the ~500-year return period math materially for a 10-year window # Directional lean per outcome - **Yes (26% Kalshi):** Southern San Andreas overdue; capable of M8.1+; UCERF3 multi-fault scenarios possible; Cascadia edge cases. **Opposing:** Zero historical precedent in 166 yrs; UCERF3 base rate only ~2%; market appears ~10x overpriced vs. science. - **No (74% Kalshi):** Strong support from scientific models, historical record, and Poisson statistics. Science implies No should be ~97–98%. # Gaps / unknowns - Whether Cascadia subduction zone offshore qualifies as "California territorial waters" (could meaningfully raise probability if so) - Exact remaining time window (market closes Dec 31, 2035 — ~10 years from ~2025) - Low market volume means 26% price may not reflect informed consensus # Calibration anchors - **Kalshi current YES: 26%** — primary anchor, but flagged as likely significantly overpriced - **UCERF3 scientific estimate: ~2–2.5%** for 10-year window - **Poisson weighted central: ~1.97%** - **Plausible range: 1.5%–5%** (science-based); market at 26% is a major outlier - Closest precedent: 1857 and 1906 events both M7.9, never reaching M8.0 threshold in recorded history

1. Both forecasters move above the UCERF3 base rate (~2%) to account for the "overdue" southern San Andreas, but neither quantifies how much the overdue narrative actually shifts the 10-year probability — the brief explicitly notes the ~500-year return period math "doesn't change materially for a 10-year window," making the upward adjustments (to 4–7%) potentially unjustified without further support. 2. Neither forecaster adequately addresses the Cascadia subduction zone edge case flagged in the brief as a genuine gap: if offshore Cascadia ruptures qualify as "California territorial waters," this could meaningfully raise the probability above UCERF3's California-only estimate, which does not appear to include Cascadia scenarios. 3. Both forecasters largely dismiss the Kalshi 26% anchor, but they differ significantly in their final estimates (4% vs. 7%) without explaining the source of this ~75% relative disagreement — the brief's plausible science-based range is 1.5–5%, suggesting Forecast 1 at 7% is already slightly outside the calibration range while Forecast 2 at 4% sits near its upper bound. 4. Neither forecaster explicitly addresses the cross-market inconsistency flagged in the brief: Kalshi's 2028 contract at 16% YES implies only ~10% incremental probability for 2028–2035, yet both forecasters treat the full 10-year window holistically without checking whether their estimates are internally consistent with that near-term implied structure. 5. Both forecasters treat the 166-year zero-event record as strong evidence, but neither notes the reference-class problem: the sample is too small to reliably distinguish between a ~0.002/yr rate and a ~0.006/yr rate, meaning the historical record provides weaker likelihood-ratio evidence than implied — the Poisson upper 95th-percentile bound (0.018/yr) is non-trivially higher than the UCERF3 central estimate.

5% (50% confidence)

Yes: 5% No: 95%

Scientific models (UCERF3, Poisson) place 10-year M8.0+ probability at ~2-2.5%, with zero such events in 166 years. The Kalshi 26% anchor is flagged as overpriced with thin liquidity and a recent -27pp correction. I land at 5%, slightly above pure base rate to account for the overdue southern San Andreas, multi-fault rupture scenarios, possible Cascadia territorial-waters edge cases, and the reference-class uncertainty noted by the critique, while staying within the science-based plausible range.

Total pipeline time: 124.7s