Scan: Will there be an at least 8.0 magnitude earthquake in California before 2028?

date	predicted	market price	confidence
2026-08-29	3%	9%	78%
2026-08-22	2%	8%	74%
2026-07-10	4%	11%	50%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: kalshi_direct kalshi_related polymarket_related wikipedia code_execution claude_news article_search

Sub-questions (Fermi decomposition)

What is the current Kalshi YES price for an M8.0+ California earthquake before 2028, and how does it compare to sibling markets (e.g., M7.0+, M7.5+) in the same series?
What does the USGS UCERF3 model estimate as the annual/30-year probability of an M8.0+ earthquake in California, and what does that imply for a Poisson probability over the remaining time until Dec 31, 2028?
Has any earthquake with an epicenter in California or its territorial waters ever reached M8.0 in the historical record (1857 Fort Tejon, 1906 San Francisco)?
Would a Cascadia Subduction Zone megathrust event count if its USGS-reported epicenter were in California waters off Humboldt/Del Norte counties, and what is the estimated probability of a Cascadia M8+ rupture in the next ~3 years?
Which faults in California are physically capable of an M8.0+ (southern San Andreas full rupture, Cascadia southern segment) and is there any recent seismic activity or scientific alert suggesting elevated near-term risk?
How does Kalshi's resolution source (USGS) determine epicenter location and magnitude, and are there ambiguities (e.g., offshore events, magnitude revisions) that could affect resolution?

Planner reasoning

This is fundamentally a base-rate question: no M8.0+ earthquake has ever been recorded with an epicenter in California (1857 Fort Tejon and 1906 San Francisco were ~7.9), and USGS UCERF3 puts the 30-year probability of an M8+ in California at roughly 7%, implying well under 1% for a ~3-year window. The main uncertainties are the market's current price (likely inflated by longshot bias), whether a Cascadia megathrust rupture off Cape Mendocino would count as 'California territorial waters', and how much time remains until close.

2. Research Parallel tool calls

tool	status	time	items	summary
kalshi_direct	OK	0.2s	1	## This Market's Kalshi Data Before 2028 (KXEARTHQUAKECALIFORNIA-28) - Current price (probability): 10.00% - 7-day price change: +1.00% - 30-day price change: -4.00% - Average daily volume: 153 contracts - Price range: 4.00% - 18.00% - Data points: 39 days
kalshi_related	OK	2.6s	2	2 related markets / summaries. series KXEARTHQUAKECALIFORNIA: 0 markets (skipped 3 no-signal) \| series KXEARTHQUAKE: 0 markets (skipped 0 no-signal) \| keyword 'earthquake California': ok \| keyword 'earthquake magnitude': ok
polymarket_related	OK	0.1s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'earthquake California': 0 markets \| keyword 'earthquake 8.0': 0 markets \| keyword 'San Andreas': 0 markets
wikipedia	OK	2.6s	5	Fetched 5 Wikipedia entries (0 missing pages).
code_execution	OK	34.8s	0	## Poisson Probability Results: M8.0+ California Earthquake Base rate derivation (UCERF3): - 7% probability over 30 years → implied annual Poisson rate λ = -ln(0.93)/30 ≈ 0.00242/yr (consistent with the ~0.0024 figure given) **Base-case probabilities (λ ≈ 0.0024/yr, no Cascadia adjustment)
claude_news	OK	22.7s	8	## Key Findings - UCERF3 (USGS official long-term forecast): The chances of earthquake magnitude 8.0 or greater hitting California in the next 30 years have been increased from about 4.7% to 7%, the U.S. Geological Survey (USGS) said . This is the base rate authorities use; scaled to a ~2-year
article_search	OK	2.6s	20	Fetched 20 unique articles across 3 queries (lookback=365d). 'California earthquake magnitude 8 probability USGS': 8 articles \| 'Cascadia subduction zone earthquake risk northern California': 8 articles \| 'San Andreas fault big one forecast': 8 articles

3. Evidence Brief Sonnet 5 · 5853 chars

# Event Will an earthquake of magnitude ≥8.0 with epicenter in California or its territorial waters occur before Dec 31, 2028? # Outcomes to forecast - Yes - No # Kalshi market anchor **Current YES price: 10%** (KXEARTHQUAKECALIFORNIA-28). 7-day change: +1pt; 30-day change: -4pt. Range over life: 4%-18%. Avg daily volume: 153 contracts over 39 data points — thin but active market. [kalshi_direct] # Sub-question answers 1. **Kalshi YES price / sibling comparison** — Current market (2028 close) = 10%. Sibling "before 2035" market = 33% (range 7%-65%, recently down 15% over 30d). No M7.0/M7.5 sibling markets found in this series. [kalshi_related] 2. **UCERF3 implied probability** — USGS UCERF3 estimates ~7% chance of M8.0+ in California over 30 years (up from 4.7% in earlier estimate), implying annual λ≈0.0024/yr. Poisson-scaled to the ~2.5-3.5yr window to end-2028: base case ~0.6%-0.84%; with generous Cascadia-spillover adjustments, ~0.8%-2.5%. [code_execution, claude_news] 3. **Historical record** — No CA-epicentered earthquake has ever recorded M8.0+; the two largest (1857 Fort Tejon, 1906 San Francisco) were both ~M7.9. [wikipedia] 4. **Cascadia counting toward CA** — Cascadia extends into northernmost CA waters near Cape Mendocino/Humboldt-Del Norte; a full-margin rupture (M8.7-9.0+) with epicenter there would plausibly count. USGS: 10-15% chance of full-margin M9 in next 50yrs; one source cites 40% chance of M8.0-8.5 for southern segment in 50yrs, scaled to ~1-2% over 2yrs — but only the far-north CA coastal sliver is CA territory, and most rupture area is offshore Oregon/Washington. [claude_news] 5. **Faults capable of M8.0+** — Southern San Andreas (unruptured >300yrs since last major event) is the most flagged, but modeled full-rupture scenarios (ShakeOut) top out ~M7.8; modeled ceiling ~M8.3 requires unusually long multi-segment rupture. 2026 Cajon Pass stress study found record 1000-year stress levels but modeled joint rupture only M7.4-7.8. No official near-term elevated-alert exists. [claude_news, article_search] 6. **USGS resolution mechanics** — Not detailed in rules; USGS assigns epicenter/magnitude via standard seismic network, with magnitude sometimes revised post-event (preliminary vs. final Mw). No explicit ambiguity documentation found for offshore/Cascadia edge cases in the market rules themselves. # Key facts (high-confidence, factual) 1. [kalshi_direct] Current YES = 10% for 2028 market; thin volume (~153/day). 2. [kalshi_related] 2035 sibling market priced at 33%, implying market believes most of the "risk mass" builds over a longer horizon. 3. [claude_news] UCERF3: 30-year M8+ probability ≈7% statewide. 4. [wikipedia] No CA earthquake has ever reached M8.0 historically (max ~7.9). 5. [claude_news] San Andreas modeled max ~M8.3, requiring an unusually long rupture beyond documented events. 6. [claude_news] Cascadia full-margin M9 event: 10-15% chance in 50 years; overlaps only marginally with CA territorial waters. 7. [claude_news] No official 2024-2026 alert of imminent M8+ risk; one non-credible blog forecast dismissed. 8. [article_search] 2026 seismicity in CA has been minor (max ~M4.8 offshore Humboldt, M4.6 Santa Cruz) — no unusual escalation signals. # Cross-market signals - Kalshi related: 2035 "before" market at 33% (vs 10% for 2028) shows term-structure consistent with roughly linear/Poisson risk accumulation over time, and implies market's own base rate differs somewhat from raw UCERF3 Poisson math (33% for +7yrs extra is higher than pure extrapolation of a ~1% 2028 rate would suggest, but could reflect illiquidity/overpricing tail risk). - Polymarket: No matching markets found (0 hits for earthquake/California/San Andreas keywords). - Sportsbook implied: N/A (not applicable to this category). # Analyst opinions and speculation - USGS/Caltech scientists flag southern San Andreas as overdue and under record stress (2026 Cajon Pass study), but their own models cap likely rupture magnitude below 8.0 (7.4-7.8). - Cascadia scientists (PNSN, Oregon officials) describe a major Cascadia event as "imminent" in generic long-term terms, but this framing is not calibrated to a 2028 deadline and mostly concerns WA/OR, not CA specifically. - No credible scientist or institution has raised near-term (2026-2028) M8+ alert specifically for California. # Directional lean per outcome - **Yes**: Weak support — nonzero physical possibility via extreme San Andreas multi-segment rupture or Cascadia southern-segment spillover into CA waters; heightened stress reported in 2026 study; market prices meaningfully above 0 (10%) reflecting some tail-risk premium. - **No**: Strong support — no CA earthquake has ever recorded M8.0+; UCERF3-derived base rates for 2028 window are <1-2.5% even with generous adjustments; current elevated-stress studies still model sub-8.0 magnitudes; recent 2025-2026 CA seismicity has been minor. # Gaps / unknowns - Precise Kalshi resolution mechanics for edge cases (offshore Cascadia epicenter attribution to "California waters," preliminary vs. revised magnitude) not documented in rules. - No quantified overlap/probability specifically for Cascadia rupture epicenter falling within CA territorial waters vs. Oregon/Washington. - Market's 33% price for the 2035 sibling appears notably higher than simple base-rate extrapolation would suggest — reason for the gap unclear (liquidity, risk premium, or extra information). # Calibration anchors - Kalshi current YES price (anchor): **10%** - UCERF3/Poisson-model base rate for ~2.5-3.5 yr window: **~0.6%-2.5%** (most likely ~1%) - Historical base rate: 0 California M8.0+ events in ~170 years of instrumental/historical record. - Sibling 2035 market: 33% (longer horizon, higher price) — useful for gauging term structure though possibly overpriced vs. physical models.

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

68% conf

Yes	3%
No	97%

Kalshi prices YES at 10%, but this is a thin, long-dated tail-risk market where lottery-ticket bias and time-value effects systematically inflate YES prices. Physical base rates are far lower: UCERF3 implies ~7% chance of M8+ in California over 30 years, which Poisson-scales to roughly 0.6%-1% for the ~2.5-3 year window to end-2028, and only ~2.5% with generous Cascadia-spillover adjustments. No California earthquake has ever registered M8.0+ (1857 and 1906 were ~M7.9), current southern San Andreas stress studies model ruptures capped at M7.4-7.8, and 2025-2026 seismicity has been unremarkable. I set YES modestly above the pure model rate to account for resolution edge cases (offshore Cascadia epicenter attribution, preliminary magnitude overestimates) but well below the market anchor.

gpt-5.6-sol

74% conf

Yes	4%
No	96%

The Kalshi YES price is 10%, but the market is thin and substantially exceeds the UCERF3-derived probability of roughly 0.6%-2.5% through 2028. No California-epicentered earthquake has historically reached M8.0, while plausible San Andreas scenarios generally top out around M7.8 and require an unusually long multi-segment rupture to exceed 8.0. Cascadia creates some additional tail risk, but only a small portion overlaps California waters and an epicenter there is unlikely. With no official near-term elevated alert or unusual seismic escalation, the evidence supports moving materially below the market anchor while retaining a nonzero extreme-event probability.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasters converge tightly (3% and 4%) far below the Kalshi anchor (10%) without adequately explaining why the market—despite being "thin"—should be discounted so heavily rather than partially trusted; thin markets can still encode information (e.g., informed traders on tail risk), and neither forecast grapples with why the 30-day trend was -4pt or what drove the 18% peak.

Neither forecast seriously engages with the term-structure anomaly: the 2035 sibling market at 33% implies a much higher effective annual rate than the 2028 market’s 10%, and if that rate were applied to a Poisson model backward, it would suggest the 2028 market itself may be underpriced at 10% relative to the longer-horizon market’s implied risk—this internal inconsistency across Kalshi markets deserves discussion rather than being waved away as "illiquidity/overpricing."

Both rely almost entirely on UCERF3/Poisson scaling and dismiss the market price, but neither considers that UCERF3's ~7%/30yr figure was revised upward from 4.7%, suggesting model uncertainty is trending toward higher tail risk, which could argue for a probability above the pure Poisson midpoint rather than converging near it.

Neither forecast explicitly addresses resolution ambiguity for preliminary vs. revised magnitude (per point 6 in the brief)—a moderately large event initially reported as ≥8.0 before downgrade could create temporary YES-triggering conditions or market disputes, a scenario that could modestly raise effective YES probability beyond pure physical occurrence odds.

Both treat the Cascadia contribution as a minor addition without quantifying it, despite the brief noting one source citing "40% chance of M8.0-8.5 for southern segment in 50 years" scaling to 1-2% over the window—this alone is comparable to or exceeds their entire point estimates, meaning the aggregate (San Andreas + Cascadia base rates) could push a more defensible range to 1.5-3% rather than clustering at 3-4%, and their tightness here appears more like anchoring to each other than fully independent uncertainty propagation.

Neither highlights that the historical zero-count base rate (0/170 years) combined with small-sample uncertainty should imply wider confidence intervals rather than tightly bunched point estimates near 3-4%, and the market's persistence in the 4-18% range over its lifetime suggests real disagreement/uncertainty that a decisive 96-97% No confidence may be understating.

6. Aggregator Fable 5.1 · final calibrated forecast

4% (71% confidence)

Yes: 4% No: 96%

Both forecasters correctly discount the 10% Kalshi anchor: this is a thin, ~3-year-dated tail market where NO holders must lock capital until 2028, so a YES price near 10% is largely consistent with time-value and lottery-ticket premia rather than genuine ~10% event odds. Physical base rates (UCERF3 ~7%/30yr Poisson-scaled to ~0.6-1%, plus a generous Cascadia southern-segment spillover of ~1-2% only partially attributable to CA waters) sum to roughly 1.5-3%, and no CA quake has ever reached M8.0 with current San Andreas stress studies still modeling M7.4-7.8 ruptures. The critique's points on term-structure inconsistency, upward UCERF3 revisions, and preliminary-magnitude resolution edge cases justify sitting at the upper end of the physical range rather than the midpoint, but not approaching the market price. I settle at 4% YES, matching the higher of the two forecasts to reflect Cascadia and resolution ambiguity.

Pipeline Timing

Total pipeline time: 131.3s

Per-tool research timings shown in the Research section above.