← Back to scans

Will there be an at least 8.0 magnitude earthquake in Japan before 2030?

KXEARTHQUAKEJAPAN-30 · Climate and Weather · 2026-04-05
43%
Multi-Step
54%
Market Price
-11.0%
Edge
45%
Confidence
Final Rationale
My base Poisson estimate using λ≈0.10/year (corrected upward for possible undercount of historical M8.0+ events) gives ~31% over 3.75 years. The Devil's Advocate correctly identified that independent source zone probabilities (Nankai Trough ~8-11% alone, plus Japan Trench, Kuril, Sagami, Ryukyu zones) should be combined multiplicatively, pushing the estimate to ~37-41%. I adjust to 43% after also accounting for broader geographic scope of 'territorial waters.' The market at 54% is higher than my corrected estimate; I attribute the ~11% gap partly to retail sentiment amplified by recent seismic activity (Jan 2025 Noto M7.6) and the market's demonstrated unreliability (historical range 28-97%), and do not fully defer to it.
What Would Change My Mind
A strong slow-slip event or seismic swarm detected along the Nankai Trough would push my estimate above 55%. Conversely, if the question's geographic scope were clarified to exclude Kuril Trench events north of Hokkaido, I'd drop to ~35%. New research revising the Nankai Trough 30-year probability significantly upward (e.g., >80%) or downward (<30%) would shift my estimate by >10 points.
1. Planning Agent Sonnet
binary science wikipedia_lookup article_search web_search kalshi_data code_execution
Sub-questions
IDQuestionWeight
sq1 Does Japan's historical seismic record show at least one M8.0+ earthquake per 4-year period on average? 45%
sq2 Are current seismic stress indicators or expert forecasts suggesting elevated probability of a major earthquake in Japan before 2030? 30%
sq3 Has there already been an M8.0+ earthquake in Japan since the market opened (resolving it early)? 25%
Reasoning Approach
Japan sits at the intersection of four tectonic plates (Pacific, Philippine Sea, Eurasian, North American), making it one of the most seismically active regions on Earth. The key analytical approach is: (1) establish a historical base rate of M8.0+ earthquakes in Japan's territory using seismological records, typically going back ~125 years of instrumental records; (2) model occurrences as a Poisson process to compute P(≥1 event) over the ~4-year window remaining until Jan 2030; (3) adjust upward for known seismic gaps and stress accumulation (the Nankai Trough is overdue for a major rupture per Japanese government forecasts, with 70-80% probability in 30 years); (4) check for any recent events that may have already resolved the market. The question differs from 'will there be any earthquake' questions because M8.0+ is a very high threshold—only a handful of Japanese earthquakes have crossed it historically. However, the Tohoku 2011 earthquake (M9.1) demonstrated that Japan can generate these events. The ~4-year window is shorter than the typical inter-event period, but Japan's elevated seismic hazard and Nankai Trough expert forecasts make this non-trivial. The Nankai Trough government forecast specifically warns of a major event, which significantly elevates priors above pure historical base rates.
2. Research Execution No LLM
ToolStatusTime
wikipedia_lookup OK 1.83s
article_search OK 0.69s
web_search OK 44.81s
kalshi_data OK 1.55s
code_execution OK 0.02s
3. Evidence Synthesis Sonnet · 10 items
#StrengthDirectionSourceClaimPriced In
1 MODERATE ↑ UP code_execution Using ~11 M8.0+ earthquakes in Japan from 1900-2025 (rate ~0.088/year), a Poisson model yields P(≥1 event in 4 years) ≈ 0.30-0.38, with a central estimate around 0.34. Yes
2 WEAK — NEUTRAL wikipedia_lookup Japan's historical earthquake list confirms recurring M8.0+ events, but the exact count and inter-event periods are difficult to establish precisely from the Wikipedia list alone, especially pre-instrumental records. Yes
3 MODERATE ↓ DOWN web_search The strongest recent earthquake near Japan in the past 10 years was M7.6 on December 8, 2025; no M8.0+ event has occurred in Japan in the past decade, suggesting inter-event gaps can exceed 10 years. Yes
4 STRONG ↑ UP web_search Japan's Earthquake Research Committee (September 2025) widened the Nankai Trough M8-9 probability to 60-94.5% within 30 years, but also offered a lower model range of 20-50%, causing public confusion. No
5 MODERATE ↑ UP article_search A December 2025 BBC article noted Japan's earthquake panel assessed a 60-90% chance of a Nankai Trough megaquake within 30 years, following a M7.5 quake in northeastern Japan that reignited concern about 'the big one'. Yes
6 MODERATE — NEUTRAL article_search The Tokara Islands (Japan) experienced over 1,800 earthquakes in three weeks as of July 2025, prompting evacuations, but the strongest was far below M8.0, indicating elevated regional seismicity without a major rupture. Yes
7 STRONG — NEUTRAL wikipedia_lookup The July 2025 Kamchatka M8.7-8.8 earthquake triggered tsunami warnings in Japan but struck in Russian territory, not Japan's; it demonstrates the seismic volatility of the broader Pacific region. Yes
8 STRONG ↓ DOWN web_search No M8.0+ earthquake has occurred in Japan as of early April 2026; the strongest recent event near Japan was M7.6 on December 8, 2025, ~80 km northeast of Hachinohe. Yes
9 STRONG ↓ DOWN article_search No news articles from the April 2025 to April 2026 search window report an M8.0+ earthquake in Japan; the major earthquake covered was the Kamchatka (Russia) M8.8 event, not in Japanese territory. Yes
10 MODERATE ↓ DOWN kalshi_data The Kalshi market price of 54% (down 2% in 7 days, up 5% in 30 days) implies the market has NOT yet resolved YES, confirming no qualifying M8.0+ Japan earthquake has occurred. Yes
Information Gaps
  • No precise enumeration of all M8.0+ Japan earthquakes since 1900 with exact dates to compute a reliable Poisson rate — the '~11 events' figure used in code execution is approximate.
  • No data on current geodetic stress accumulation along the Nankai Trough or other Japanese fault systems (e.g., GPS uplift rates, slow-slip event frequencies) that could refine near-term probability.
  • No breakdown of what time window remains until January 1, 2030 (approximately 3.75 years from April 2026), which affects lambda calculation.
  • The two conflicting Nankai Trough probability models (60-90% vs 20-50% over 30 years) have not been reconciled into a clear near-term (4-year) conditional probability.
  • No expert seismological assessment specifically addressing the probability of M8.0+ in Japan within the 2026-2030 window (as opposed to the standard 30-year horizon).
  • No data on whether recent seismic swarms (Tokara Islands, December 2025 M7.6 near Hachinohe) reflect stress loading or release that changes near-term M8.0+ probability.
Key Uncertainties
  • Whether the Nankai Trough will rupture before 2030: expert forecasts (60-90% over 30 years) provide weak constraint on the 4-year window but suggest elevated hazard.
  • Accuracy of the historical M8.0+ base rate for Japan — the count of ~11 events since 1900 may undercount or overcount depending on definitional boundaries of 'Japan territory'.
  • Whether the 2025 Kamchatka M8.8 earthquake relieved or increased stress on adjacent Japan fault systems.
  • The two conflicting Japanese government probability models for Nankai create genuine scientific uncertainty about hazard level.
  • Whether current Kalshi market pricing (54%) accurately reflects all available seismological information or is influenced by retail speculation (wide price range of 28-97% suggests high uncertainty).
Market Context
Price: 0.54 · 7d: -0.02 · 30d: +0.05
Volume: LOW-MODERATE — 120 contracts/day average · Depth: UNKNOWN — not provided
Reliability: LOW-MEDIUM — moderate volume but extremely wide historical price range (28-97%) suggests thin market susceptible to sentiment swings rather than informed pricing; 170-day data history is reasonable but retail-driven
4. Independent Forecaster Opus · No market price Did NOT see market price
Independent estimate: 35% (45% confidence)
sq1: Does Japan's historical seismic record show at least one M8.0+ earthquake per 4-year period on average?
32%
Base rate: Using ~11 M8.0+ earthquakes in Japan from 1900-2025 (125 years), the annual rate λ ≈ 0.088/year. For a Poisson process, P(≥1 in 3.75 years) = 1 - e^(-0.088*3.75) = 1 - e^(-0.33) ≈ 0.28. However, the count of '~11' is approximate. If we use a slightly higher count (say 12-13 events including borderline cases in territorial waters), λ could be ~0.096-0.104, giving P ≈ 0.30-0.32. The decade-long gap since Tohoku 2011 (no M8.0+ in Japan for ~15 years) pushes slightly down, but earthquakes are memoryless in Poisson models so this shouldn't affect the calculation. The moderate evidence about the M7.6 in December 2025 being the strongest in a decade confirms inter-event times can be long. I'll estimate this at 0.32, reflecting the Poisson calculation with some uncertainty in the exact count.
sq2: Are current seismic stress indicators or expert forecasts suggesting elevated probability of a major earthquake in Japan before 2030?
55%
The Japanese government's Earthquake Research Committee forecasts 60-94.5% probability of a Nankai Trough M8-9 within 30 years (with a lower model range of 20-50%). Taking the midpoint of the broader range (~60% over 30 years), the annualized rate would be λ_Nankai ≈ -ln(1-0.60)/30 ≈ 0.031/year, giving P(≥1 Nankai event in 3.75 years) ≈ 11%. Using the higher estimate (90% in 30 years), λ ≈ 0.077/year, P ≈ 25%. But Nankai is only ONE source zone — Japan has multiple M8.0+ capable zones (Japan Trench, Kuril Trench, Sagami Trough, etc.). The Nankai-specific probability adds to the baseline from other zones. The elevated recent seismicity (Tokara swarm, M7.6 near Hachinohe) provides weak upward pressure but could equally represent stress release. The Kamchatka M8.8 is neutral — it's in Russian territory but shows regional plate boundary activity. Overall, expert forecasts do suggest elevated hazard compared to a naive base rate, pushing the answer toward 'yes, there is elevated probability.' I interpret this sub-question as asking whether conditions are elevated above the historical base rate, and the answer is moderately yes. P_yes = 0.55.
sq3: Has there already been an M8.0+ earthquake in Japan since the market opened (resolving it early)?
2%
Strong evidence from multiple recent searches confirms no M8.0+ earthquake has occurred in Japan as of early April 2026. The strongest recent event was M7.6 on December 8, 2025. This is a near-certain 'no' — the only residual uncertainty is whether an event occurred in the last few days that hasn't been picked up by the search results, which is extremely unlikely for an M8.0+ event (it would dominate global news). P_yes = 0.02.
Show full reasoning trace
base rate: Japan has experienced roughly 11 M8.0+ earthquakes from 1900-2025, giving an annual rate of ~0.088. Over the remaining ~3.75 years (April 2026 to January 2030), a Poisson model gives P(≥1) ≈ 28-32%. This is the anchor estimate.
evidence updates: Expert forecasts for the Nankai Trough (60-94.5% over 30 years per Japan's Earthquake Research Committee) push the estimate UP by ~3-6%, since the Nankai Trough may be in a more stressed state than the average historical period, though much of this is already captured in the base rate. The 15-year gap since the last M8.0+ in Japan (Tohoku 2011) is neutral in a Poisson framework but could suggest slight stress accumulation, pushing UP by ~1-2%. The absence of any M8.0+ so far in the question period pushes DOWN slightly (~1%) since about 0.7 years of the window have passed without an event. Recent seismic swarms are too weak to meaningfully adjust. Net adjustment: +3-5% above pure Poisson base rate.
combination method: I primarily rely on the Poisson base rate (~30%) with modest upward adjustment from expert Nankai Trough forecasts and stress considerations, yielding ~35%. The early resolution check (sq3) confirms no event yet, so the full forward-looking probability applies to the remaining ~3.75 years.
final: My integrated estimate is 35% probability of an M8.0+ earthquake in Japan before January 1, 2030. This reflects the historical base rate of ~0.088/year applied to the remaining window, with modest upward adjustment for Nankai Trough stress indicators and expert forecasts. The confidence is moderate (0.45) due to genuine scientific uncertainty about the exact base rate, the conflicting Nankai probability models, and the inherent unpredictability of earthquake timing.
5. Devil's Advocate Sonnet Market price revealed
Math Error Reasoning Flaw Missing Info
Challenges
  1. The base rate calculation of '~11 M8.0+ earthquakes in Japan from 1900-2025' is acknowledged as approximate but may significantly undercount events. The USGS catalog shows Japan-region M8.0+ events including 1923 Kanto (7.9-8.2 depending on source), 1933 Sanriku (8.4), 1944 Tonankai (8.1), 1946 Nankai (8.1), 1952 Tokachi-oki (8.1), 1958 Etorofu (8.3), 1963 Kuril Islands (8.5), 1968 Tokachi-oki (8.2), 1994 Kuril (8.3), 2003 Tokachi-oki (8.3), 2011 Tohoku (9.1), plus several borderline events. Depending on how 'Japan' is defined geographically (especially including Kuril Trench events in Japan's EEZ), the count could be 13-16+, which would push λ to 0.10-0.13/year and P(≥1 in 3.75 years) to 31-39%.
  2. The sub-question decomposition is somewhat redundant and the combination method is unclear. Sq1 (historical base rate), sq2 (current stress indicators), and sq3 (already resolved?) don't combine multiplicatively or additively in any standard way. The forecaster essentially just uses the base rate with a modest adjustment, making sq2's p_yes of 0.55 largely decorative — it's unclear how this specific number feeds into the final 35%.
  3. The forecaster treats the 15-year gap since Tohoku 2011 as 'neutral in a Poisson framework' but then also mentions it 'could suggest slight stress accumulation.' These are contradictory — either the process is Poisson (memoryless) or it's not. If there's a renewal process component (stress accumulation), the gap should provide a more meaningful upward adjustment than the +1-2% assigned.
  4. The Nankai Trough probability is underweighted. The forecaster converts the 60% over 30 years to ~11% over 3.75 years for Nankai alone, then notes this is 'only ONE source zone.' But the forecaster doesn't properly add the Nankai-specific increment to the base rate from OTHER zones. If Nankai alone contributes ~11-25% (depending on which model), and other zones contribute their own independent probabilities, the combined probability should be calculated as 1 - (1-P_Nankai)(1-P_other), which would yield meaningfully higher than the 30-35% estimated.
  5. The forecaster notes the information gap about the exact definition of 'Japan' for resolution purposes. An M8.0+ earthquake on the Japan Trench, Kuril Trench near Hokkaido, or Ryukyu Trench could all potentially count. The geographic scope of what qualifies likely broadens the probability beyond the most conservative interpretation.
  6. The confidence level of 0.45 seems appropriate given the uncertainties, but the point estimate of 35% may be too anchored to the simple Poisson calculation without fully incorporating the independent Nankai Trough hazard. A proper multi-source-zone calculation would likely yield 35-45%.
Suggested adjustment: +8pp
The primary reasoning flaw is the failure to properly combine independent source zone probabilities. The Nankai Trough alone has an annualized ~11-25% probability over 3.75 years (depending on model), and this should be combined with the probability from OTHER source zones (Japan Trench, Kuril Trench, Sagami Trough, etc.) using 1-(1-P_Nankai)(1-P_other). Additionally, the base rate count of ~11 events may undercount, with a more careful enumeration potentially yielding 13-16 events and a higher λ. These two independent reasoning issues both push the estimate upward by roughly 5-10%. Adjusting from 0.35 to 0.43.
Market comparison: Forecaster: 35%, Market: 54%. The 19-point divergence is partially explained by genuine reasoning flaws (improper combination of independent source zone probabilities, possible undercount of historical events). However, the market's low-medium reliability (thin, retail-driven, with an enormous historical range of 28-97%) means the market price itself carries limited informational value. A corrected estimate of ~43% narrows the gap but doesn't fully close it, which is appropriate given the market's unreliability.
6. Final Forecast Opus
43% (45% confidence)
My base Poisson estimate using λ≈0.10/year (corrected upward for possible undercount of historical M8.0+ events) gives ~31% over 3.75 years. The Devil's Advocate correctly identified that independent source zone probabilities (Nankai Trough ~8-11% alone, plus Japan Trench, Kuril, Sagami, Ryukyu zones) should be combined multiplicatively, pushing the estimate to ~37-41%. I adjust to 43% after also accounting for broader geographic scope of 'territorial waters.' The market at 54% is higher than my corrected estimate; I attribute the ~11% gap partly to retail sentiment amplified by recent seismic activity (Jan 2025 Noto M7.6) and the market's demonstrated unreliability (historical range 28-97%), and do not fully defer to it.
7. Entry Decision Opus Trader
Side
NO
Position Size
$800
Contracts
1739
Trader Confidence
45%
Trade Rationale
The forecaster estimates 43% probability of an M8.0+ earthquake in Japan before 2030, while the market prices this at 54%. This yields an 11% edge on the NO side. The reasoning is grounded in Poisson modeling calibrated against historical seismicity rates and independent zone probabilities. The Devil's Advocate flagged potential math errors and suggested an upward adjustment of +8%, which the forecaster already incorporated into the 43% figure. The 1366-day horizon introduces substantial uncertainty, and the forecaster confidence is moderate at 0.45, reflecting genuine uncertainty about seismic forecasting. However, the edge is meaningful at 11% and the direction is clear.
Allocation Logic
With an 11% edge but only 0.45 forecaster confidence, a long time horizon (3.75 years), and low-moderate liquidity, I size below the $1000 baseline at $800. The long resolution time and inherent unpredictability of earthquakes warrant caution despite the decent edge.
Entry price: $0.46
Current: $0.40
Status: OPEN
P&L: -$104.35
Pipeline Timing
StepTime
Planning 28.1s
Research 44.81s
Synthesis 35.1s
Forecaster 37.77s
Critic 26.35s
Final 24.01s