← Back to scans

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

KXEARTHQUAKECALIFORNIA-28 · Climate and Weather · 2026-04-22
2%
Agent
11%
Market Price
-9.0%
Edge
78%
Confidence
Volume: 6,014
Spread: 5.0c
Days to resolution: 983
Markets in event: 1
Final Rationale
UCERF3-derived Poisson calculations give ~1.2-1.5% probability of M8.0+ in California over this 2.7-year window, with historical recurrence intervals of 100-200 years supporting this range. I partially accept the Devil's Advocate critique that historical magnitude uncertainty (1857 Fort Tejon, 1906 SF both ~M7.9 with ±0.2-0.3 measurement error) and model uncertainty justify a slight upward adjustment to 2%. I reject the market signal of 11% because the orderbook is demonstrably broken ($0.98 spread, one-sided book, ~138 contracts/day, bids at $0.01), which reflects retail disaster-overestimation bias rather than informed pricing — this is a textbook unreliable market. The 9-point divergence from market is justified by robust seismological consensus and the market's documented illiquidity.
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
1. Planner Opus · Fermi decomposition + tool plan
Tools chosen: kalshi_data kalshi_orderbook wikipedia_lookup article_search web_search code_execution
Sub-questions (Fermi decomposition)
  1. Is California's base rate seismic hazard (from UCERF3 and historical recurrence data) consistent with a probability above 2% for an M8.0+ event occurring within the ~2.7 year resolution window? (w=0.50)
  2. Has there been any recent increase in seismic activity or precursory signals along California's major fault systems (San Andreas, Cascadia, Hayward) that would suggest elevated near-term risk above baseline? (w=0.30)
  3. Does the current Kalshi prediction market price for this event reflect a probability broadly consistent with seismological base rate calculations for an M8.0+ earthquake in California before 2028? (w=0.20)
Planner reasoning
This question requires seismic hazard analysis using the Poisson process framework standard in earthquake science. The key reference class is the USGS Uniform California Earthquake Rupture Forecast (UCERF3), which gives fault-specific M8.0+ probabilities over 30-year windows. The resolution window is ~2.7 years (983 days), which is short relative to geological recurrence intervals for M8.0+ events. California's primary M8.0+ sources are: (1) full San Andreas rupture similar to 1857 Fort Tejon event, (2) Cascadia subduction zone offshore NorCal/OR, and (3) combinations of other faults. UCERF3 estimates roughly 7% chance of M8.0+ on the San Andreas in 30 years, implying ~0.24% annual probability from that source alone. Historical record since 1700 in California shows perhaps 2-3 events arguably qualifying as M8.0+ (1700 Cascadia, 1857, 1906), suggesting recurrence of 100-200 years. Temporal analysis: for YES to resolve, a single large rupture event must occur before Dec 31, 2028 with no warning possible given the nature of earthquake forecasting. There are no precursory events that reliably predict M8.0+ within years. The base rate calculation suggests a probability in the 1-3% range for this window. The Kalshi market price will serve as an important cross-check. Key differentiator from other natural disaster questions: earthquake timing is essentially memoryless (Poisson), so elevated recent activity provides only marginal signal.
2. Research Parallel tool calls · legacy v2
ToolStatusTimeItemsSummary
kalshi_data OK 2.0s -
kalshi_orderbook OK 0.2s -
wikipedia_lookup OK 1.9s -
article_search OK 0.2s -
web_search OK 53.7s -
code_execution OK 0.0s -
3. Evidence Synthesis Sonnet · 12 items · legacy v2
#StrengthCredibilityDirectionSourceClaimPriced In
1 STRONG 82 ↓ DOWN code_execution Poisson process calculations using UCERF3-implied annual probabilities yield a statewide combined probability of ~1.45% for an M8.0+ earthquake in California within the ~2.7-year window; a composite weighted estimate lands at ~1.20%. Yes
2 MODERATE 75 ↑ UP wikipedia_lookup The San Andreas Fault is capable of large strike-slip earthquakes (e.g., 1857 Fort Tejon M7.9), and the Cascadia subduction zone offshore NorCal can produce M9.0+ events; both are recognized M8.0+ sources for California. Yes
3 MODERATE 72 ↑ UP wikipedia_lookup The Cascadia subduction zone extends into Northern California and is capable of producing 9.0+ magnitude earthquakes, representing a significant but geographically marginal contribution to California-epicenter M8.0+ probability. Yes
4 STRONG 80 ↓ DOWN code_execution Using a 100-year recurrence interval (optimistic/short), probability is ~2.66%; using 200-year recurrence (more realistic), it falls to ~1.34%; UCERF3-implied San Andreas-only probability is only ~0.65% over 2.7 years. Yes
5 MODERATE 68 NEUTRAL web_search A M4.6 earthquake struck near Boulder Creek/Santa Cruz Mountains on April 2, 2026 — the strongest quake in this part of the US in over 2 years — but aftershock forecasts show <1% chance of M5+ follow-on and no indication of imminent large rupture. No
6 WEAK 55 NEUTRAL web_search A M3.6 earthquake and a swarm of minor quakes occurred near San Ramon/Dublin (East Bay/Hayward Fault region) in early February 2026, but no significant damage or indication of escalating seismic hazard was reported. Yes
7 MODERATE 60 ↓ DOWN web_search No confirmed precursory signals or anomalous seismic activity along the San Andreas, Cascadia, or Hayward faults suggestive of elevated near-term M8.0+ risk were found in recent reporting; standard earthquake science confirms such signals are not reliably predictive anyway. Yes
8 WEAK 50 NEUTRAL web_search California's 2026 seismic hazard context notes 16 million people in extreme hazard zones and ongoing infrastructure vulnerability, but this reflects chronic baseline risk, not an acute near-term signal. Yes
9 STRONG 65 ↓ DOWN kalshi_data The Kalshi market for M8.0+ California earthquake before 2028 is currently priced at 11%, substantially above the seismological base rate estimate of ~1.2–1.5%, suggesting significant market overpricing relative to scientific consensus. No
10 MODERATE 60 ↓ DOWN kalshi_data The 7-day price change for the 2028 market is -6 percentage points (from ~17% to 11%), and the 30-day change is +1%, indicating recent downward volatility; the wide historical price range (2%–22%) suggests this market has been poorly calibrated or driven by sentiment. No
11 STRONG 70 ↓ DOWN kalshi_orderbook The orderbook shows a bid-ask spread of $0.98 on YES/NO sides (yes_bid=$0.01, no_bid=$0.01), indicating extremely thin liquidity on YES side despite high stated depth (9030 contracts), raising serious questions about the reliability of the 11% headline price. No
12 MODERATE 58 ↓ DOWN kalshi_data The related market for M8.0+ California before 2035 is priced at 28% (down sharply from 73% range), and Japan M8.0+ before 2030 is at 50%; these comparators suggest the California 2028 market may have recently corrected toward more realistic levels but remains elevated vs. base rates. No
Information Gaps
  • No direct UCERF3 report data accessed — relied on secondary computations; ideal to have the specific UCERF3 Table 1 M8.0+ probability figures for all California fault sources combined
  • No data on Cascadia subduction zone-specific recurrence interval updates post-2015 (e.g., paleoseismology studies that may have refined the ~300-500 year recurrence estimate)
  • No USGS ShakeAlert or recent ANSS seismic catalog data to confirm or deny elevated microseismicity along specific fault segments
  • The article_search returned 0 results, leaving a gap in recent (April 2025–April 2026) scientific or news reporting on California seismic hazard updates
  • No information on whether the Kalshi market has been subject to manipulation, wash trading, or low-information retail participation that could explain the wide gap between market price and base rates
  • No data on strain accumulation rates or GPS geodetic measurements along the San Andreas, which could refine near-term hazard estimates
Key Uncertainties
  • Whether the Cascadia subduction zone's epicenter boundary (near Cape Mendocino) would qualify events as 'California epicenter' per resolution criteria — this could significantly change the probability calculation
  • The recurrence interval assumption for California M8.0+ is highly uncertain (ranges from ~100 to 300+ years depending on source combination), directly driving the 0.89%–2.66% probability range
  • Earthquake occurrence is memoryless (Poisson), so even a very recent large event would not reduce risk; conversely, absence of recent events does not increase it — this limits the value of precursory signal analysis
  • Whether the Kalshi market price reflects informed participants with private seismological models, or simply reflects retail sentiment and low-information trading given poor orderbook depth
  • Potential for simultaneous multi-fault rupture scenarios (e.g., Southern San Andreas + adjacent faults) that might not be fully captured in single-fault UCERF3 estimates
Market Context
Price: 0.11 · 7d: -0.06 · 30d: +0.01
Volume: LOW — 138 avg daily contracts is very low for a binary event market · Depth: MISLEADING — stated depth 9030 but bid prices at $0.01 suggest the book is one-sided and not actionable; the spread of $0.98 indicates near-zero real liquidity on YES
Reliability: LOW — wide historical price range (2%–22%), low volume, anomalous orderbook structure, and large gap vs. scientific base rates all suggest this market price is unreliable as a probability signal
4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price
Independent estimate: 2% (82% confidence)
base rate
2%
82% conf
evidence driven
2%
85% conf
contrarian
2%
78% conf
Spread
0pp
Low disagreement
sq1: Is California's base rate seismic hazard consistent with >2% probability for M8.0+ in ~2.7 year window?
20%
Multiple Poisson calculations converge on 1.2-1.5% for this window, with only the optimistic 100-year recurrence assumption pushing above 2% (2.66%). UCERF3 San Andreas-only gives 0.65%. Historical record since 1700 (1700 Cascadia, 1857 Fort Tejon - though technically M7.9, 1906 - also ~M7.9) shows M8.0+ strictly in California is rare; arguably zero pure California M8.0 events in 300+ years. The question asks if base rate EXCEEDS 2%, which requires optimistic assumptions. P ~ 20%.
sq2: Recent precursory signals elevating near-term risk?
8%
No credible precursory signals identified. M4.6 Santa Cruz and M3.6 San Ramon events are within normal background seismicity. Standard seismology: such signals are not reliably predictive. Earthquakes are memoryless. Minor swarms occur routinely without escalation. P ~ 8%.
sq3: Does Kalshi market price reflect seismological base rate?
40%
No direct evidence on current market price. Prediction markets on rare natural disasters often overprice due to retail sentiment and low liquidity, but Kalshi markets with seismology-informed traders may converge near base rates. Without data, assume moderate probability of alignment. P ~ 40%.
Show full reasoning trace
base rate: UCERF3-derived Poisson calculations give ~1.2-1.5% for M8.0+ in California over 2.7 years. Historical frequency since 1700 is roughly 2-3 qualifying events (and several of those were M7.9, not strictly M8.0+), implying ~100-200 year recurrence, which translates to 1.3-2.6% for this window.
evidence updates: No precursory signals or elevated activity warrant deviation from base rate (near-zero update). Earthquake Poisson memorylessness means recent quiescence or activity provides minimal signal. Cascadia contribution is marginal for California-epicenter events. Keep estimate anchored near 1.3-1.5%.
combination method: Rather than mechanically combining sub-question probabilities (which ask diagnostic meta-questions), I anchor directly to the seismological base rate since that's the dominant evidence. The event itself is what's being forecast.
final: Final estimate: 1.5% — slightly above central Poisson base rate to account for (a) Cascadia boundary ambiguity, (b) multi-fault rupture scenarios not fully captured, and (c) general model uncertainty. High confidence in the low probability given robust seismological consensus.
ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.015, 'confidence': 0.82}, 'evidence_driven': {'p_yes': 0.015, 'confidence': 0.85}, 'contrarian': {'p_yes': 0.02, 'confidence': 0.78}}, 'spread': 0.005, 'individual_p_yes': {'base_rate': 0.015, 'evidence_driven': 0.015, 'contrarian': 0.02}}
5. Devil's Advocate Sonnet · legacy v2 Market price revealed
Overconfidence
Challenges
  1. The forecaster's base rate calculation may be too narrowly anchored on 'strictly California M8.0+' events. UCERF3 actually estimates ~7% probability of M8.0+ in California over 30 years, which prorates to ~0.65% per year or ~1.75% over 2.7 years — but this is a point estimate with substantial model uncertainty. The forecaster settles at 1.5% which is reasonable, but the 'high confidence' (0.817) may be overstated given acknowledged gaps in direct UCERF3 access and Cascadia recurrence updates.
  2. Sub-question 3 (market alignment) was given 40% weight conceptually but then essentially discarded in the combination_method, which states the forecaster anchors directly to seismological base rate. This is actually defensible here given the market's low liquidity and wide spread, but the sub-question structure was essentially vestigial — worth noting the decomposition didn't drive the final answer.
  3. The forecaster correctly identifies that the Kalshi market is illiquid (bid-ask $0.98, 138 contracts/day, one-sided orderbook) and therefore unreliable. This justifies not moving toward 11%. The market price likely reflects retail overestimation of disaster probabilities, a well-documented bias in low-liquidity natural disaster markets.
  4. Minor concern: the forecaster notes 1857 Fort Tejon was M7.9 and 1906 was ~M7.9, treating these as sub-threshold. However, magnitude estimates for historical events have uncertainty of ±0.2-0.3, so some of these could qualify as M8.0+ depending on the measurement. This slight upward tail is partially captured in the 1.5% vs 1.3% adjustment but could justify slightly more.
Suggested adjustment: +0pp
A very small upward adjustment (to ~2%) is warranted because: (1) historical magnitude uncertainty means some 'M7.9' events might qualify, (2) the forecaster's 0.817 confidence seems high given acknowledged information gaps on UCERF3 specifics and Cascadia updates, and (3) multi-fault rupture scenarios and tail risk from model uncertainty favor slight conservative widening. The forecaster's rejection of the 11% market price is well-justified given documented low liquidity and one-sided orderbook.
Market comparison: Forecaster: 1.7%, Market: 11%. The divergence is large but the market signal is unreliable (low volume ~138/day, $0.98 bid-ask spread, one-sided orderbook, wide historical range 2-22%). Scientific base rates from UCERF3 strongly support a sub-3% estimate. The forecaster correctly identified that this market likely reflects retail disaster-overestimation bias rather than informed pricing. No adjustment toward market is warranted.
6. Aggregator Opus · final calibrated forecast
2% (78% confidence)
UCERF3-derived Poisson calculations give ~1.2-1.5% probability of M8.0+ in California over this 2.7-year window, with historical recurrence intervals of 100-200 years supporting this range. I partially accept the Devil's Advocate critique that historical magnitude uncertainty (1857 Fort Tejon, 1906 SF both ~M7.9 with ±0.2-0.3 measurement error) and model uncertainty justify a slight upward adjustment to 2%. I reject the market signal of 11% because the orderbook is demonstrably broken ($0.98 spread, one-sided book, ~138 contracts/day, bids at $0.01), which reflects retail disaster-overestimation bias rather than informed pricing — this is a textbook unreliable market. The 9-point divergence from market is justified by robust seismological consensus and the market's documented illiquidity.
7. Entry Decision Opus Trader
Side
NO
Position Size
$1200
Contracts
1348
Trader Confidence
78%
Trade Rationale
The forecaster's 2% estimate is well-grounded in UCERF3 Poisson base rates with ~100-200 year recurrence intervals, giving a solid 9-point edge on NO. Forecaster confidence is high at 0.78, and the reasoning is scientifically robust rather than speculative. The market's 11% price clearly reflects retail disaster-overestimation bias on an illiquid book (~138 contracts/day), which is a classic exploitable inefficiency in tail-risk prediction markets. The long 983-day horizon is already accounted for in the base-rate calculation, and there's no portfolio concentration risk — no other seismic/disaster positions open.
Allocation Logic
Sized above baseline ($1200) due to strong edge (9pts) and high forecaster confidence (0.78) on a well-understood statistical base rate, but capped below $1500 due to low liquidity, long time-to-resolution (2.7 years of carry risk), and the small but real tail risk of an actual M8.0+ event.
Entry price: $0.89
Current: $0.89
Status: OPEN
P&L: $0.00