Tools chosen: kalshi_direct wikipedia claude_news code_execution kalshi_related polymarket_related

Sub-questions (Fermi decomposition)

Planner reasoning

This is a rare-event geological question with a fixed deadline (end of 2028). The key driver is the base rate of M8.0+ earthquakes in California, which is extremely low historically. I'll anchor on the Kalshi market price, gather scientific/USGS background on California seismic hazard, and compute a base-rate probability over the roughly 3-year window.

# Current state The market resolves YES if a M8.0+ earthquake occurs with epicenter in California or its territorial waters before Dec 31, 2028. No such event has occurred in California's instrumental record (~124 years); the most recent large events were the Dec 2024 M7.0 Cape Mendocino and historical ~M7.9 events in 1857/1906. The question is whether one occurs in the ~2.5 remaining years of the window. # Timeline of key events - **1857-01-09**: Fort Tejon earthquake, est. M7.9 on San Andreas — confirmed (pre-instrumental) - **1906-04-18**: San Francisco earthquake, est. M7.9 — confirmed - **2014**: UCERF3 published; ~7% statewide M8.0+ probability over 30 years — confirmed - **2024-12-05**: M7.0 offshore Cape Mendocino; noted to slightly increase stress on northern San Andreas — confirmed [USGS] - **2024**: Dr. Lucy Jones notes "2024 had more earthquakes than any year since 1988" — reported [KTLA] - **2025 (recent)**: Research finds southern San Andreas + San Jacinto faults at "critically loaded" stress levels, highest in ~1,000 years — reported [BBC Science Focus] - **2026-05 (recent)**: New research: Cascadia and San Andreas can "sync up" within minutes/hours; 3 instances in last 1,500 years — reported [ScienceDaily] --- # Event Will a M8.0+ earthquake occur with epicenter in California or territorial waters before Dec 31, 2028? # Outcomes to forecast - Yes - No # Kalshi market anchor **Current YES price: 11.00%** (KXEARTHQUAKECALIFORNIA-28) - 7-day change: **−7.00%** (recent downward move) - 30-day change: +1.00% - Price range (28 days): 5%–18% - Avg daily volume: 108 contracts - **Related:** KXEARTHQUAKECALIFORNIA-35 (before 2035) trades at **36%** — implies ~25 percentage points of additional probability for years 2028–2035, roughly consistent with UCERF3 rates. The 2028 market at 11% appears **significantly overpriced** vs. scientific base rates (~0.7–1.5%). # Sub-question answers 1. **Kalshi implied probability for M8.0+ CA before 2028?** — 11% YES, down 7 points over 7 days. Range 5–18% over 28 days. [Kalshi direct] 2. **Historical base rate of M8.0+ in California?** — Zero confirmed M8.0+ events in ~124 years of instrumental record. Fort Tejon (1857) and SF 1906 both estimated ~M7.9. MLE rate = 0/yr; fault-model rate ~0.0024/yr. [Wikipedia, code_execution] 3. **Can CA faults produce M8.0+, and USGS probability?** — Yes; San Andreas could produce ~M8.0 but most scenario ruptures cap ~7.8–7.9. UCERF3 (2014) gives ~7% statewide M8+ over 30 years → ~0.72% over 3 years. Southern CA specifically: ~4.7% over 30 yrs → ~0.48% over 3 yrs. [USGS/UCERF3, code_execution] 4. **Recent seismic activity or heightened risk?** — Dec 2024 M7.0 Cape Mendocino (slightly loads northern San Andreas); Southern San Andreas + San Jacinto at "historically high" stress (1,000-year peak per 2025 research); new Cascadia-San Andreas synchronization risk identified. [USGS, BBC Science Focus, ScienceDaily] 5. **Probability of M8.0+ in ~3-year window?** — UCERF3-derived: ~0.72%; Gutenberg-Richter upper bound: ~2.4%; Bayesian (zero-data adjusted): ~1.2%. Best estimate: **~0.7–1.5%**. [code_execution] # Key facts (high-confidence, factual) 1. [UCERF3/USGS] Statewide M8.0+ probability: ~7% over 30 years → ~0.0024/yr annual rate 2. [Wikipedia/USGS] No M8.0+ event instrumentally recorded in California in 124 years 3. [code_execution] Poisson 3-year probability using UCERF3 rate: **~0.72%** 4. [USGS] Dec 2024 M7.0 Cape Mendocino confirmed; slightly increases stress on northern San Andreas 5. [BBC Science Focus] Southern San Andreas + San Jacinto faults at highest stress in ~1,000 years (2025 research) 6. [Wikipedia] San Andreas Fault: ~750 mi long, 20–35mm/yr slip rate; produces M~8 class events historically # Cross-market signals - **Kalshi KXEARTHQUAKECALIFORNIA-35 (before 2035):** 36% YES — implies ~10-year window; the ratio 11%/36% is internally inconsistent with flat annual rates, suggesting market mispricing or different trader pools - **Kalshi KXEARTHQUAKEJAPAN-30 (before 2030):** 58% YES — reflects Japan's much higher M8+ frequency; useful calibration contrast - **Polymarket:** No active California earthquake market found # Analyst opinions and speculation - Dr. Lucy Jones (Caltech): 2024 had highest earthquake activity since 1988 [KTLA] - USGS Elizabeth Cochran: Southern CA has 36% chance M7.5+ in next 30 years [Newsweek] - New 2026 research: Cascadia-San Andreas synchronization could compound disaster risk, but historically rare (3 instances in 1,500 years) [ScienceDaily] - Cajon Pass "earthquake gate" research: If both southern San Andreas and San Jacinto rupture simultaneously, magnitude could exceed standalone estimates [BBC Science Focus] # Directional lean per outcome - **Yes (M8.0+ before 2028):** - Supporting: Historically high fault stress; new multi-fault rupture science; 2024 seismic uptick; M8+ is physically possible on San Andreas - Opposing: Zero M8.0+ in 124-year record; 3-year window is very short vs. recurrence intervals (~200–500 yrs); UCERF3 anchors at ~0.7%; most SA scenarios top at ~7.8–7.9 - **No:** - Strong support: Scientific consensus ~0.7–1.5% probability; no instrumental precedent; short window; Kalshi's 11% appears 7–15x above base rate # Gaps / unknowns - Whether "territorial waters" epicenter materially expands probability (offshore Mendocino/Cascadia proximity) - Exact remaining time window (market closes Dec 31, 2028 — currently ~2.5 years remain) - Why Kalshi market sits at 11% vs. ~1% scientific estimate — possible thin market / retail speculation # Calibration anchors - **Kalshi current YES price: 11%** — PRIMARY anchor, but appears significantly overpriced - **UCERF3 scientific base rate: ~0.7%** for 3-year window — strongest scientific anchor - **Plausible range: 0.7%–2.4%** (UCERF3 to G-R upper bound) - **Precedent:** No M8.0+ in California in recorded history; Fort Tejon/1906 both ~M7.9; Japan M8+ recurrence ~decades vs. California ~centuries

1. Both forecasters dismiss the Kalshi 11% anchor too aggressively without adequately explaining *why* it is wrong beyond labeling it "retail speculation." The brief notes the 28-day range is 5–18% and volume is thin (~108 contracts/day), meaning the price reflects weak price discovery — but neither forecaster engages with whether thin-market overpricing justifies landing at 2.5–3% rather than, say, 5%, given genuine model uncertainty. 2. Neither forecaster quantifies the impact of the "territorial waters" inclusion on probability. The brief flags this as a gap, noting offshore Mendocino/Cascadia proximity could materially expand the effective fault zone covered — the Cascadia subduction zone produces M8+ events on a ~200–500 year recurrence cycle and the 2026 synchronization research explicitly links it to the San Andreas, yet both forecasters treat this as a minor footnote. 3. Both forecasters anchor heavily on UCERF3's ~0.72% figure but fail to critically examine whether UCERF3 adequately incorporates the specific 2025 findings — southern San Andreas + San Jacinto at "highest stress in ~1,000 years" — which the brief presents as potentially outside the model's calibration window, making the UCERF3 rate a possible underestimate for this specific near-term period. 4. The internal Kalshi cross-market inconsistency (11% for 2028 vs. 36% for 2035) is noted in the brief as evidence of mispricing, but neither forecaster uses this as a quantitative check: if 36% over ~10 years is roughly consistent with UCERF3, then the residual 2028–2035 probability should be ~25%, implying the 2028 window should be roughly ~7–8% — closer to the market price and higher than both forecasters' estimates. 5. Both forecasters treat the M8.0 threshold as clearly distinct from M7.9 historical events, but the brief explicitly notes that 1857 and 1906 are "estimated ~M7.9" with inherent uncertainty in pre-instrumental measurements — a point that cuts both ways on whether there is truly "zero precedent," yet neither forecaster addresses magnitude measurement uncertainty as a source of epistemic risk. 6. The forecasters' final estimates of 2.5–3% represent a 2–4x uplift over the UCERF3 base rate (~0.72%), but neither provides a principled decomposition of what drives that uplift (model uncertainty? stress loading? offshore expansion? synchronization risk?), making it difficult to assess whether 2.5–3% is well-calibrated or itself arbitrary.

4% (50% confidence)

Yes: 4% No: 96%

Scientific base rates (UCERF3 ~0.7%, upper bound ~2.4%) strongly favor No over the short ~2.5-year window with zero M8.0+ in California's instrumental record. Both forecasters landed near 2.5-3%; the critique fairly notes elevated fault stress, territorial-waters/offshore inclusion, and the cross-market implied ~7-8% deserve some weight. I nudge slightly upward to 4% to reflect genuine model uncertainty and thin-market signal, while staying well below Kalshi's 11%.

Total pipeline time: 156.2s