Tools chosen: kalshi_direct article_search claude_news kalshi_related polymarket_related code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

planner fallback: broad sweep + category conditionals

# Current state No M8.0+ earthquake has occurred in California since at least 1906. The resolution criterion requires an epicenter in California or its territorial waters at ≥M8.0 before Dec 31, 2035. The market is currently open with ~10 years remaining in the window. # Timeline of key events - **1906-04-18**: San Francisco earthquake (~M7.7–7.9) — last event approaching M8.0 in California. [USGS, confirmed] - **2014**: UCERF3 published; estimates ~7% M8.0+ probability over 30 years statewide. [USGS, confirmed] - **2024-12-05**: M7.0 earthquake recorded off California coast near San Andreas Fault; no M8.0+ followed. [GeoQuake.org, confirmed] - **2025 (current)**: Kalshi market at 36%; +10% in 7 days, +7% in 30 days. [Kalshi, confirmed] --- # Event Will there be an M8.0+ earthquake in California (or territorial waters) before Dec 31, 2035? # Outcomes to forecast - Yes (M8.0+ occurs before 2035) - No (no M8.0+ before 2035) # Kalshi market anchor **Current YES price: 36%** — significantly elevated vs. scientific base rates. Trend: +10% over 7 days, +7% over 30 days (sharp recent rise). Price range historically 9%–65%; current 36% sits in upper-mid range. Volume modest at 67 contracts/day, suggesting limited liquidity discipline. **The 36% price appears to significantly overestimate scientific consensus.** # Sub-question answers 1. **Current Kalshi price for KXEARTHQUAKECALIFORNIA-35?** — 36% YES, up +10% in 7 days and +7% in 30 days. [Kalshi direct] 2. **Recent news affecting the event?** — Dec 2024 M7.0 off California coast drew attention but no M8+ followed. Southern San Andreas shows strain accumulation but no imminent precursor signals detected. [GeoQuake.org, OctagonAI] 3. **Related prediction markets?** — Manifold implied ~8% for M8+ before 2030. OctagonAI model estimates ~24% for before 2035 vs. 41% market price (suggesting overpricing). No directly comparable Polymarket market found. [OctagonAI] 4. **Historical base rate?** — UCERF3 puts M8.0+ at ~7% over 30 years (~2014 baseline), implying ~2.3% over 10 years. Historical record (1 event/200–300 yrs) implies 1.6%–4.9% over 10 years. Gutenberg-Richter extrapolation outlier suggests ~18%, but seismologists prefer characteristic-fault model. [USGS UCERF3, code analysis] # Key facts (high-confidence, factual) 1. [USGS UCERF3] M8.0+ probability in California over 30 years ≈ 7%; annual rate ≈ 0.00233/yr 2. [USGS UCERF3] Increase from UCERF2 (~4.7%) driven by multi-fault rupture scenarios 3. [Code analysis] UCERF3-implied 10-year probability (2025–2035): ~2.3%; Poisson process is memoryless so no conditional uplift from 1906 elapsed time 4. [Historical record] No confirmed M8.0+ in California proper since 1906 (~119 years) 5. [Code analysis] Historical base rate range across scenarios: 1.6%–4.9% for 10-year window 6. [CRMP] M6.7+ has >99% chance in 30 years; M8.0+ threshold is far more demanding 7. [OctagonAI] Southern San Andreas Coachella segment shows substantial slip deficit but no deep non-volcanic tremors detected # Cross-market signals - **Kalshi (this market):** 36% YES — sharply above scientific estimates; recent +10% 7-day spike unexplained by news - **Polymarket:** No directly relevant M8.0+ California earthquake market found - **OctagonAI/Manifold:** ~8% (before 2030), ~24% model estimate (before 2035) — both well below current Kalshi price # Analyst opinions and speculation - OctagonAI model explicitly flags the Kalshi market as likely overpriced (model: 24.2% vs. market: ~41% at time of writing). [OctagonAI] - USGS: "Tectonic forces are continually tightening the springs of the San Andreas fault system, making big quakes inevitable" — but this refers to geological timescales, not a near-term signal. [USGS] - Seismologists favor characteristic earthquake model over Gutenberg-Richter for M8+ estimation, supporting lower tail probabilities. [code analysis] # Directional lean per outcome - **Yes (M8.0+ before 2035):** - Supporting: 10-year window provides meaningful exposure; Southern San Andreas strain accumulation; multi-fault rupture scenarios in UCERF3; Dec 2024 M7.0 shows fault activity - Opposing: UCERF3 base rate only ~2–3% for 10 years; no M8.0+ in 119 years; no imminent precursor signals; characteristic earthquake model suppresses G-R extrapolation - **No:** - Supporting: Scientific consensus strongly favors No (~97% under UCERF3 10-yr estimate); historical rarity; Poisson memorylessness eliminates "overdue" uplift - Opposing: Long window; accumulated tectonic strain; territorial waters inclusion expands risk area slightly # Gaps / unknowns - Reason for sharp +10% 7-day price spike on Kalshi is unexplained (thin market, noise?) - UCERF3 is from 2014; UCERF4 may revise estimates - "Territorial waters" scope could include Cascadia-adjacent offshore zones, modestly expanding risk - Market liquidity is low (67 contracts/day), making price less reliable as consensus signal # Calibration anchors - **Kalshi current YES price: 36%** (primary anchor, but flagged as likely overpriced) - **UCERF3 scientific base rate: ~2.3% for 10-year window** (strongest quantitative anchor) - **OctagonAI model: ~24%** (intermediate; methodology unclear) - **Manifold (before 2030): ~8%** (shorter window, consistent with ~3–5% annualized) - Best-estimate synthesis: **3–8% YES**, with the Kalshi 36% appearing to reflect thin-market noise or uninformed speculation rather than calibrated probability

1. Both forecasters land at exactly 8% Yes, which is suspiciously round and suggests anchoring on the Manifold "before 2030" figure rather than an independent synthesis; neither explains why 8% is preferable to, say, 5% or 12%, given the genuine uncertainty range of 2.3%–24% across cited sources. 2. Both forecasters largely dismiss the OctagonAI ~24% estimate without engaging its methodology; the brief notes it is an explicit model estimate (not thin-market noise), and simply labeling its methodology "unclear" is insufficient justification for discounting it to near-zero weight relative to UCERF3. 3. Neither forecaster adequately addresses the territorial-waters scope expansion, which the brief flags as potentially including Cascadia-adjacent offshore zones — a non-trivial addition given that the Cascadia subduction zone produces M8+ events and sits within plausible territorial-water boundaries of California. 4. Both treat the Kalshi 36% price almost entirely as noise to be dismissed, but the brief notes prices have historically ranged 9%–65% and the market has been open long enough to accumulate some informed trading; a more rigorous critique would require showing the 36% is structurally mispriced rather than simply asserting thin-market noise explains a +10% move. 5. The "Poisson memorylessness" point is invoked to eliminate "overdue" uplift, but UCERF3 itself uses a time-dependent (BPT) model for characteristic faults precisely because elapsed time since last rupture does matter for specific fault segments like the Southern San Andreas Coachella section — neither forecaster distinguishes between the statewide Poisson aggregate and the fault-specific time-dependent hazard. 6. Neither forecaster considers the possibility that UCERF4 revisions (flagged as a gap in the brief) could substantially alter base rates upward, which represents a genuine tail risk to the "scientific consensus strongly favors No" framing both adopt.

10% (50% confidence)

Yes: 10% No: 90%

Scientific base rate (UCERF3) implies ~2.3% over the 10-year window, while model estimates range up to ~24% (OctagonAI) and the Kalshi market sits at 36% in a thin, noisy market. The critique fairly notes both forecasters clustered at exactly 8% and underweighted the OctagonAI model, territorial-waters scope, and fault-specific time-dependent hazard. I nudge slightly above their 8% to 10% to account for the long window, model uncertainty, and boundary/scope factors, while still discounting the illiquid Kalshi price as overpriced relative to seismological consensus.

Total pipeline time: 165.1s