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

← Back to scans

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

KXEARTHQUAKECALIFORNIA-28 · Climate and Weather · 2026-05-02

Agent

11%

Market Price

-7.5%

Edge

70%

Confidence

Volume: 6,016

Spread: 5.0c

Days to resolution: 973

Markets in event: 1

Final Rationale

UCERF3 implies ~0.6-1% probability of M≥8.0 in California over this 2.66-year window, and zero confirmed M≥8.0 events have occurred in California's instrumental history (1906 SF and 1857 Fort Tejon both ~M7.9). I PARTIALLY ACCEPT the critique on ensemble correlation, magnitude-threshold uncertainty around 8.0 (1906 estimates span 7.7-8.3), and Cascadia/territorial-waters ambiguity — these structural uncertainties justify lifting from 2% to ~3.5%. I REJECT deferring to the 11% market price: liquidity is extremely thin ($0.98 spread, ~123 daily contracts), pricing reflects retail disaster-salience rather than informed seismic forecasting, and 11% is 5-17x scientific base rates with no precursor evidence. My 7.5pp deviation from market is justified by the explicit UCERF3 Poisson rate plus negligible historical realization.

Price vs. prediction

How the market moved after the agent's call. Solid blue line is the actual market price; dashed line is the agent's static forecast at $t{=}0$.

$n = 58$ follow-ups

Re-scan Context

This market has been scanned before. Previous predictions:

Date	Predicted	Market Price	Confidence
2026-04-22	2%	11%	78%

1. Planner Opus · Fermi decomposition + tool plan

Tools chosen: kalshi_data kalshi_orderbook wikipedia_lookup article_search code_execution

Sub-questions (Fermi decomposition)

Is there a historical base rate supporting a ≥8.0 earthquake in California occurring within any given ~3-year window? (w=0.50)
Are there current seismic precursors, fault stress indicators, or scientific forecasts that suggest elevated probability of a ≥8.0 earthquake in California before 2028? (w=0.30)
Does the UCERF3 or comparable authoritative seismic hazard model assign a meaningful (>5%) probability to a ≥8.0 event in California within a 3-year window? (w=0.20)

Planner reasoning

This question requires estimating the probability of a rare, high-magnitude (≥8.0) earthquake in California within a specific ~2.67-year window (May 2026 – Dec 2028). Key considerations: 1. **Reference class / base rates**: M≥8.0 earthquakes in California are extremely rare. The UCERF3 model (the authoritative source) estimates ~7% probability over 30 years for M≥8.0 in California, implying a ~0.24% annual rate. Historical records suggest 0-2 events in ~200 years of documented seismicity (1857 Fort Tejon ~7.9, 1906 San Francisco ~7.9 are close but may not strictly qualify as 8.0). This converts to roughly 0.6-1.5% probability over 2.67 years. 2. **Poisson process assumption**: Earthquakes are well-modeled as Poisson processes in the absence of known precursors. There's no memory effect — a fault that 'hasn't ruptured in a long time' does increase hazard per elastic rebound theory, but this is already incorporated into UCERF3. 3. **What makes this question hard**: M8.0 is a very specific threshold — the 1906 earthquake was estimated at M7.9 by modern standards, which would NOT resolve YES. The San Andreas, Cascadia (partially in California territorial waters), and offshore faults are the primary candidates. The Cascadia subduction zone is the most capable of M8+ but its epicenter is largely in Oregon/Washington. 4. **Temporal analysis**: For YES to occur in 973 days, a major fault rupture would need to initiate within California or its territorial waters. No credible short-term prediction exists. The timeframe is too short for significant secular hazard accumulation beyond background rates. Scientists cannot predict individual earthquakes; only probabilistic assessments apply. 5. **Market calibration**: The Kalshi market price will serve as an important anchor. Given the ~0.6-2% estimated probability range from models, the market should be pricing this in low single digits.

2. Research Parallel tool calls · legacy v2

Tool	Status	Time	Items
kalshi_data	OK	2.2s	-
kalshi_orderbook	OK	0.2s	-
wikipedia_lookup	OK	1.8s	-
article_search	OK	0.7s	-
code_execution	OK	0.0s	-

3. Evidence Synthesis Sonnet · 12 items · legacy v2

#	Strength	Credibility	Direction	Source	Claim	Priced In
1	STRONG	82	↓ DOWN	code_execution	UCERF3-based calculation gives ~0.64% probability of M≥8.0 in California over 2.66 years; historical base rate approach (using ~0-2 events in 200 years) gives ~2.63%; midpoint is ~1.64%.	Yes
2	STRONG	88	↓ DOWN	wikipedia_lookup	The 1906 San Francisco earthquake has an estimated moment magnitude of 7.9, meaning it would NOT meet the ≥8.0 threshold — reducing the historical count of qualifying events in California to near zero.	Yes
3	STRONG	90	NEUTRAL	wikipedia_lookup	UCERF3 is the authoritative official seismic hazard model for California, providing probabilistic estimates of earthquake rupture likelihood; it is used for engineering design, building codes, and insurance.	Yes
4	WEAK	55	NEUTRAL	article_search	A 4.9-magnitude earthquake struck near Indio Hills, Southern California (near the San Andreas fault) in January 2026, but this is routine seismicity far below M8.0 threshold and indicates no elevated hazard signal.	Yes
5	MODERATE	65	↓ DOWN	article_search	No articles found indicating current scientific warnings, elevated stress signals, or precursors specifically pointing to an imminent M≥8.0 earthquake in California before 2028.	Yes
6	WEAK	50	NEUTRAL	article_search	An 8.8-magnitude earthquake struck Russia's Kamchatka Peninsula in July 2025, generating tsunami warnings and waves reaching California's coast — but the epicenter was in Russia, not California, and does not indicate elevated California fault stress.	Yes
7	STRONG	85	↓ DOWN	code_execution	UCERF3 implies roughly 7% probability of M≥8.0 in California over 30 years (~0.242% annual rate), which converts to only ~0.64% over a 2.66-year window — well below the 5% threshold mentioned in the sub-question.	Yes
8	STRONG	88	↓ DOWN	wikipedia_lookup	UCERF3 is the latest (2015) authoritative ERF for California but does not provide a specific >5% probability estimate for M≥8.0 within any 3-year window; all evidence points to sub-5% for this timeframe.	Yes
9	MODERATE	55	↑ UP	kalshi_data	The Kalshi market for M≥8.0 California earthquake before 2028 is currently priced at 11%, which is significantly above the UCERF3-derived 0.64–2.63% range, suggesting possible overpricing or that market participants are using a different (higher) base rate assumption.	No
10	MODERATE	50	↑ UP	kalshi_data	The Kalshi market for M≥8.0 California before 2035 is at 34%, which implies roughly 3% per year — substantially higher than UCERF3's 0.24%/year, suggesting the market may be using inflated priors or conflating M≥7.5 with M≥8.0.	Yes
11	MODERATE	60	NEUTRAL	kalshi_orderbook	The orderbook shows extremely wide spread ($0.98) with yes_bid and no_bid both at $0.01 and depth of 10,207, indicating very thin active trading; the 11% price may not reflect genuine informed market consensus.	No
12	WEAK	40	↓ DOWN	kalshi_data	The California before 2028 market dropped 6% in the last 7 days, suggesting recent selling pressure that may reflect a correction toward more scientifically grounded estimates.	No

Information Gaps

No current USGS or UCERF3 annual update or any 2025-2026 revised hazard assessment for California M≥8.0 specifically was found in the research data
No data on current GPS-measured fault creep rates, interseismic coupling, or stress accumulation on the southern San Andreas or offshore faults that would update the UCERF3 baseline
No information on whether the Cascadia subduction zone's northernmost extent reaches California territorial waters (which could be the only realistic pathway to M≥8.0 with a California epicenter)
No expert commentary or scientific papers specifically discussing the short-term (2025–2028) probability of a M≥8.0 California event
Lack of clarity on whether 'territorial waters' in the resolution criteria would include the offshore Cascadia zone segment near the CA-OR border
No seismological community statements or USGS advisory notes suggesting any elevated short-term hazard for California M≥8.0

Key Uncertainties

Whether the resolution criteria's 'territorial waters' would include significant portions of the Cascadia subduction zone, which is by far the most capable fault of generating M≥8.0 near California
The true historical base rate is highly uncertain given the short (~200 year) instrumental record; the 1857 Fort Tejon earthquake (~M7.9) and 1906 San Francisco (~M7.9) both may not qualify, leaving zero confirmed M≥8.0 California events on record
Whether Kalshi market pricing at 11% reflects genuine informed probability assessment or retail participant overestimation driven by availability bias (large earthquakes are salient)
The Poisson assumption may understate tail risk if fault stress has accumulated substantially since the last major rupture (elastic rebound), but this is already incorporated into UCERF3
Any undiscovered or unmapped offshore fault capable of M≥8.0 with epicenter in California territorial waters

Market Context

Price: 0.11 · 7d: -0.06 · 30d: +0.00
Volume: LOW · Depth: SHALLOW — despite high listed depth (10,207), the $0.98 spread with both yes_bid and no_bid at $0.01 indicates no active market-making
Reliability: LOW — extremely wide spread, very thin active trading, 123 avg daily contracts; price likely reflects retail speculation rather than informed probabilistic forecasting. Market price of 11% is ~5-17x higher than UCERF3/scientific estimates of 0.64-2.63%, a major red flag for reliability.

4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price

Independent estimate: 2% (81% confidence)

base rate

80% conf

evidence driven

82% conf

contrarian

82% conf

Spread

0pp

Low disagreement

sq1: Is there a historical base rate supporting a ≥8.0 earthquake in California occurring within any given ~3-year window?

Base rate from UCERF3: ~0.64% over 2.66 years. Historical approach: 0-1 qualifying events in 200 years gives ~0-1.3%. The 1906 SF (M7.9) and 1857 Fort Tejon (M7.9) both fall just below the 8.0 threshold. Midpoint estimate ~1.5%.

sq2: Are there current seismic precursors or forecasts suggesting elevated probability?

No credible short-term precursors exist. No scientific consensus or USGS advisory points to elevated near-term M8+ hazard. Routine M4-5 events near San Andreas don't shift the probability meaningfully. Stays near base rate ~1.5%.

sq3: Does UCERF3 assign >5% probability to M≥8.0 within 3-year window?

UCERF3 implies ~0.64% over 2.66 years, well below 5%. The sub-question asks if the model assigns >5%, which it clearly does not. Probability of YES on this sub-question is very low ~2%.

Show full reasoning trace

base rate: UCERF3 gives ~7% for M≥8.0 over 30 years in California, implying ~0.64% over 2.66 years. Historical record shows zero confirmed M≥8.0 events in California (1857 Fort Tejon and 1906 SF both ~M7.9). Base rate is roughly 0.5-2%.

evidence updates: No precursor evidence elevates the probability. Recent foreign quakes (Kamchatka 8.8) are irrelevant to California fault stress. No USGS short-term elevated hazard advisories. All evidence keeps estimate near base rate.

combination method: Weighted average across three highly correlated sub-questions all pointing to low probability. Final ~1.6%, rounded to 2% for tail-risk uncertainty.

final: Probability of M≥8.0 California earthquake before end of 2028 is approximately 2%, anchored to UCERF3 Poisson rates with small adjustment for model uncertainty and territorial waters ambiguity.

ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.02, 'confidence': 0.8}, 'evidence_driven': {'p_yes': 0.02, 'confidence': 0.82}, 'contrarian': {'p_yes': 0.02, 'confidence': 0.82}}, 'spread': 0.0, 'individual_p_yes': {'base_rate': 0.02, 'evidence_driven': 0.02, 'contrarian': 0.02}}

5. Devil's Advocate Sonnet · legacy v2 Market price revealed

Overconfidence Missing Info Reasoning Flaw

Challenges

The three sub-questions are not independent — they all essentially ask the same thing (what does UCERF3 say?). Averaging three highly correlated estimates creates false confidence (spread = 0.0 across ensemble is a red flag for groupthink, not robustness).
The forecaster may be too literal about the 8.0 threshold. The 1906 SF quake is variously estimated between M7.7 and M8.3 depending on methodology; magnitude estimates have inherent uncertainty of ±0.2-0.3. A future quake near M7.9-8.0 could plausibly be classified either way, which adds tail probability the forecast doesn't capture.
The resolution criteria ambiguity around territorial waters and the Cascadia subduction zone segment near CA-OR border is acknowledged as an information gap but not quantitatively incorporated. Cascadia can produce M9+ events, and if a Cascadia rupture's epicenter falls within CA territorial waters, this materially changes the probability.
UCERF3 is from 2015 — over a decade old. The forecaster doesn't account for model uncertainty: Poisson assumptions may understate clustering/triggering effects, and time-dependent models (BPT) can give meaningfully different short-term probabilities depending on time-since-last-event on key faults.
Confidence of 0.813 on a 973-day-out tail-risk forecast seems overstated given the multiple acknowledged information gaps (no current GPS/strain data, no updated hazard assessment, ambiguous resolution criteria).

Suggested adjustment: +2pp

Modest upward adjustment (from 2% to ~3.5%) is warranted not because of market price, but because: (1) the three sub-questions are highly correlated and shouldn't compound certainty, (2) magnitude measurement uncertainty around the 8.0 threshold adds tail probability (1906 SF is borderline), (3) Cascadia territorial-waters ambiguity adds non-trivial probability mass, and (4) UCERF3 Poisson model uncertainty itself. The market at 11% is likely too high (thin/retail-driven), but 2% is too confident given these structural uncertainties.

Market comparison: Forecaster: 2%, Market: 11%. The market is shallow with $0.98 spread and low volume — likely retail speculation overweighting salient disaster risk. The scientific base rate strongly supports a low probability. However, the forecaster's 2% may itself be slightly overconfident given measurement-threshold ambiguity, resolution-criteria ambiguity (Cascadia/territorial waters), and model uncertainty in UCERF3. A reasoned estimate is probably 3-5%, still well below market.

6. Aggregator Opus · final calibrated forecast

4% (70% confidence)