Tools chosen: kalshi_direct kalshi_related polymarket_related wikipedia claude_news code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a long-horizon geophysical base-rate question. The dominant driver is the scientific hazard estimate: USGS/UCERF3 models put the 30-year probability of M8.0+ in California at roughly 7%, implying a much lower ~10-year probability. Research should anchor on the Kalshi price, pull the scientific base rates, check cross-market signals, and compute the Poisson-scaled probability for the ~10-year window to 2035.

# Current state Kalshi's KXEARTHQUAKECALIFORNIA-35 currently prices YES at 25%, down sharply from recent highs (7-day: -6pts, 30-day: -18pts), implying the market has repriced lower after an initial spike. Resolution requires an actual M8.0+ earthquake epicentered in California/territorial waters before 2035-12-31 — a physical event, not a forecast or poll, and no such event has occurred yet. # Timeline of key events - **1906-04-18** (confirmed): San Francisco earthquake, M7.9 — largest historical CA quake, below the M8.0 threshold. [Wikipedia] - **2015** (confirmed): USGS/CGS/SCEC/CEA release UCERF3, raising the 30-year statewide M8.0+ probability from 4.7% (UCERF2) to 7.0%, citing multi-fault rupture scenarios. [conservation.ca.gov, southern.scec.org] - **No update since 2015** (confirmed by absence): No newer official USGS/UCERF numeric revision to the 7% figure has been found. [claude_news] - **2025/2026** (reported): Burkhard et al. (JGR: Solid Earth) publish 1,000-year stress modeling of the Cajon Pass "earthquake gate" (southern San Andreas/San Jacinto junction), finding stress at a millennium high — but modeling a joint rupture of ~M7.4–7.8, still below M8.0. [CNN, ScienceFocus, June 2026 reporting] - **Recent weeks** (market data): Kalshi YES price fell from as high as 65% (52-week range) to 25% now, a -18pt move in 30 days — direction suggests fading of an earlier speculative spike (possibly tied to the Cajon Pass coverage) rather than new negative evidence. # Event Will an earthquake of magnitude ≥8.0 with epicenter in California (or territorial waters) occur before Dec 31, 2035? # Outcomes to forecast - Yes - No # Kalshi market anchor **Current YES price: 25%** (KXEARTHQUAKECALIFORNIA-35). Trend: -6pts over 7 days, -18pts over 30 days; range has been 7%–65% over 51 data points — high volatility, likely driven by news cycles (e.g., Cajon Pass study) rather than fundamentals. Average daily volume ~125 contracts (thin/illiquid). Related shorter-horizon market KXEARTHQUAKECALIFORNIA-28 (before 2028) prices YES at only 11%, up modestly (+6pts/30-day). # Sub-question answers 1. **Kalshi price/trend**: 25% now, down from a 65% peak; falling 6-18pts over recent weeks — market cooling off after a spike. [kalshi_direct] 2. **UCERF3 30-year probability**: ~7.0% statewide for M8.0+ (up from 4.7% under UCERF2); scaling to a ~10-year window via Poisson gives ~2.4%. [conservation.ca.gov, code_execution] 3. **Historical record**: Zero confirmed M8.0+ CA epicenters in the ~170-year documented record; the largest (1906, M7.9) falls just short. Treating 1906 as a near-miss data point implies recurrence intervals of 150-300 years and a 10-year probability of ~3-6% — roughly double UCERF3's model-based estimate. [Wikipedia, code_execution] 4. **San Andreas capability**: Physically capable of M8.0-8.3 via multi-segment rupture (southern segment most likely). New 2025/26 stress modeling shows the Cajon Pass junction at a 1,000-year stress high ("critically loaded"), but the authors' own scenario estimate (M7.4-7.8) falls below the M8.0 threshold and explicitly does not predict timing. [southern.scec.org, CNN/ScienceFocus] 5. **Related markets**: Kalshi's own shorter-horizon "before 2028" market prices only 11% YES; a Japan M8.0-before-2030 market prices 38%. No matching Polymarket markets found (0 matches). [kalshi_related, polymarket_related] 6. **Recent forecast updates**: No formal USGS/UCERF probability update since 2015 (still 7%/30yr). The 2025/26 Cajon Pass paper is a stress-modeling study, not a revised numeric forecast, and its magnitude estimate is sub-M8. [claude_news] # Key facts (high-confidence, factual) 1. [conservation.ca.gov/UCERF3] Statewide 30-year M8.0+ probability = 7.0%. 2. [Wikipedia] No confirmed M8.0+ CA epicenter earthquake in documented history (1906 was M7.9). 3. [CNN/ScienceFocus 2025-26] Southern San Andreas/San Jacinto junction stress at 1,000-year high per new physics-based study; modeled joint rupture ~M7.4-7.8, below threshold. 4. [kalshi_direct] Current YES = 25%, down 18pts in 30 days from a 65% high. # Cross-market signals - Kalshi related: "Before 2028" variant prices only 11% YES (much lower, consistent with shorter horizon = lower cumulative risk). - Kalshi related: Japan M8.0-before-2030 prices 38% — higher, reflecting Japan's more active subduction-zone seismicity vs. CA's strike-slip regime. - Polymarket: No matching markets exist for triangulation. # Analyst opinions and speculation - Outside geologist (Weingarten, via CNN): stress state is a "1,000-year high," but this is about a sub-M8 scenario and explicitly not a timing prediction. - Model-based (UCERF3) vs. historical-analogy estimates diverge (2.4% vs. 3-6% over 10 years), reflecting genuine scientific uncertainty on recurrence. # Directional lean per outcome - **Yes**: Elevated stress findings (2025-26) and historical-analogy math (3-6%/10yr) could support a probability above UCERF3's baseline; market's recent 65% peak suggests some speculative belief in near-term risk. - **No**: Zero historical M8.0+ CA epicenters ever recorded; UCERF3 physically-based model gives only ~2.4%/10yr; the most cited recent "elevated risk" study still estimates sub-M8 magnitudes. Market has already fallen 18pts, suggesting overreaction is being corrected downward. Base rates strongly favor No. # Gaps / unknowns - No updated USGS numeric M8+ probability since 2015; unclear how the Cajon Pass stress findings would formally update UCERF3 if incorporated. - Ambiguity on whether 1857/1906 quakes count as "near M8" for base-rate purposes. - Market liquidity is thin (~125 contracts/day), so 25% may not reflect deep, informed pricing. # Calibration anchors - Kalshi current YES price (anchor): **25%** - UCERF3 model-implied 10-year probability: **~2.4%** - Historical-analogy implied 10-year probability: **~3-6%** - Kalshi's own shorter-horizon (2028) market: **11%** - Reasonable synthesized band: **~5-15%**, well below current 25% market price, suggesting Kalshi may still be overpriced despite recent declines.

1. Both forecasts converge on identical 8% figures with nearly identical reasoning, essentially just averaging the UCERF3 (2.4%) and historical-analogy (3-6%) bands without independently stress-testing whether the Poisson scaling from a 30-year to 10-year window is even appropriate for large, aperiodic ruptures—recurrence models for M8+ events are often not memoryless, so this scaling could understate true probability. 2. Neither forecaster grapples with the "reasonable synthesized band" explicitly given in the brief as ~5-15%, and both land at the very bottom of that range (8%) without justifying why they discount the upper half, especially given the Cajon Pass "1,000-year stress high" finding, which they treat as negligible despite explicit uncertainty about how such stress modeling would update UCERF3 if incorporated. 3. Both forecasts anchor heavily on the current 11% price for the "before 2028" market as corroboration, but this is a shorter horizon and should mechanically be lower than the 2035 market—using it as validation for an 8% figure (which is barely above the shorter-horizon price) may understate the extra ~7 years of exposure this longer-dated contract carries. 4. Neither analysis addresses tail/model-risk scenarios: e.g., multi-segment or triggered ruptures (South Andreas + San Jacinto + Cascadia interactions), or a paradigm-shifting reassessment of the 1857/1906 events as underestimated in magnitude due to instrumental limitations, which could push the historical base rate meaningfully higher. 5. Both forecasters treat the Kalshi price collapse (65%→25%) purely as a "fading speculative spike," but don't consider the alternative explanation that the market may retain information not yet reflected in USGS's stale (2015) baseline—if the market is aggregating diffuse expert sentiment about elevated risk, arriving at identical 8% estimates from both models risks anchoring too rigidly on the same "scientific" reference class while dismissing all price signal. 6. The near-total agreement between the two independent forecasts (both exactly 0.08/0.92, same confidence 0.72) is itself suspicious for "independent" analyses—it may reflect shared reliance on the same brief numbers rather than genuinely independent reasoning, understating the true uncertainty range that should exist given "genuine scientific uncertainty on recurrence" explicitly noted in the brief.

10% (72% confidence)

Yes: 10% No: 90%

Both forecasts landed at 8%, well-grounded in UCERF3 (~2.4%/10yr) and historical analogy (3-6%/10yr), and correctly discount the thin, news-driven Kalshi price of 25% that has already collapsed from a speculative 65% peak. However, the critique lands two valid points: the shorter-horizon 2028 market at 11% mechanically implies the 2035 contract should carry more premium than 8%, and Poisson scaling may understate risk if the southern San Andreas is late in its cycle with stress at a 1,000-year high. The Cajon Pass study's own magnitude estimate (M7.4-7.8) remains sub-threshold, so it doesn't justify a large upgrade, but model uncertainty about multi-fault ruptures warrants sitting slightly higher within the brief's synthesized 5-15% band. I move to 10% YES — above the pure scientific baseline, still far below the illiquid market anchor.

Total pipeline time: 100.3s