← Back to scans

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

KXEARTHQUAKEJAPAN-30 · Climate and Weather · 2026-05-15
40%
Agent
58%
Market Price
-18.0%
Edge
45%
Confidence
Volume: 24,090
Spread: 2.0c
Days to resolution: 1326
Markets in event: 1
Final Rationale
My independent Poisson base rate analysis (~30% for the 3.63-year remaining window using ~0.079 M8+/year for Japan) combined with source-specific Nankai/Japan Trench hazard estimates yielded 33.7%. I partially accept the Devil's Advocate critique: the combination method did underweight sq2's more-informed source-specific estimate, and JMA's tendency to report higher magnitudes than USGS for Japanese events (demonstrated by the April 2026 Sanriku M7.7 vs M7.4 discrepancy) asymmetrically favors higher probability if Kalshi resolves on JMA. I reject the implication of moving to market: the 58% price reflects thin liquidity (118 contracts/day) and likely overweights Nankai megaquake advisory headlines, and the historical 126-year catalog doesn't support that level. I deviate 18 points below market because the naive Poisson rate is ~31% and even generous source-summation across Nankai+Japan Trench+Kuril rarely exceeds ~45% for a <4-year window.
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:
DatePredictedMarket PriceConfidence
2026-05-08 48% 58% 40%
2026-05-01 42% 66% 40%
2026-04-05 43% 54% 45%
1. Planner Opus · Fermi decomposition + tool plan
Tools chosen: wikipedia_lookup article_search web_search kalshi_data code_execution
Sub-questions (Fermi decomposition)
  1. Is Japan's historical base rate of M8.0+ earthquakes consistent with at least one occurring in any given ~4-year window? (w=0.45)
  2. Are current seismic indicators (stress accumulation, recent seismic activity) elevated above baseline, suggesting heightened near-term risk of a M8.0+ event in Japan? (w=0.30)
  3. Has a M8.0+ earthquake already occurred in Japan or its territorial waters since the market opened? (w=0.25)
Planner reasoning
This question requires combining historical base rate analysis with current seismic science. The key reference class is: how frequently do M8.0+ earthquakes occur in Japan or Japanese territorial waters? Japan sits at the intersection of four tectonic plates (Pacific, Philippine Sea, Eurasian, North American) making it one of the world's most seismically active regions. Historical USGS/JMA records since ~1900 show roughly 8-12 M8+ events in or near Japan over ~125 years, yielding a Poisson rate of approximately 0.07-0.10 per year. Over a ~3.6-year window (May 2026 to Jan 2030), this translates to a meaningful probability. The critical additional factor is the Nankai Trough: Japanese government scientists estimate a 70-80% probability of a M8-9 Nankai Trough megathrust earthquake within 30 years, with elevated near-term concern. This systematic stress accumulation raises the prior above naive historical rates. The temporal analysis: ~1326 days remain (~3.6 years). For YES, a single M8+ event anywhere in Japanese territory suffices. Japan averages roughly one M8+ event per decade in conservative estimates. The Poisson model suggests roughly 25-40% probability. The Nankai Trough warning, recent seismic activity patterns, and the broad geographic scope of 'Japan or territorial waters' push this higher. I'll compare against Kalshi market pricing to calibrate.
2. Research Parallel tool calls · legacy v2
ToolStatusTimeItemsSummary
wikipedia_lookup OK 2.0s -
article_search OK 0.6s -
web_search OK 48.3s -
kalshi_data OK 3.7s -
code_execution OK 0.0s -
3. Evidence Synthesis Sonnet · 12 items · legacy v2
#StrengthCredibilityDirectionSourceClaimPriced In
1 STRONG 75 ↑ UP code_execution Based on ~10 M8+ events in Japan over 126 years (rate ~0.079/year), the Poisson probability of at least one M8+ event in the remaining ~4.63-year window is approximately 30.8%. Yes
2 WEAK 55 NEUTRAL wikipedia_lookup Japan's earthquake list confirms it is one of the world's most seismically active regions with frequent major earthquakes, though the Wikipedia data did not provide a precise count of M8+ events. Yes
3 MODERATE 72 ↑ UP web_search Japan accounts for ~10% of the world's earthquakes and ~20% of M6+ earthquakes globally, confirming extremely high seismic activity in the region. Yes
4 STRONG 85 ↑ UP article_search Japan's earthquake investigation panel (as of September 2025) estimated a 60-90% chance of a Nankai Trough megaquake (M8-9) within the next 30 years, indicating elevated long-term stress accumulation. Yes
5 MODERATE 80 NEUTRAL article_search A M7.5 earthquake struck northeastern Japan in December 2025, triggering a megaquake advisory; officials stated only a ~1% chance of M8+ in the following week, and no megaquake followed. Yes
6 STRONG 82 ↑ UP web_search A M7.4-7.7 earthquake struck off Honshu on April 20, 2026 (Sanriku), prompting officials to warn of a 1% chance of M8+ in the following week (10x higher than normal baseline), but no M8+ event materialized. Yes
7 MODERATE 78 ↑ UP article_search Japan issued a megaquake advisory estimating potential tsunami heights of up to 98 feet (~30m) and nearly 200,000 potential deaths, underscoring ongoing elevated risk concern from authorities. Yes
8 WEAK 40 NEUTRAL article_search Wikipedia's slow earthquake article notes that slow/silent earthquakes and tremor are detectable along subduction zones, which can be precursors to stress accumulation; however no specific Japan slow-quake data was retrieved. Yes
9 STRONG 85 ↓ DOWN web_search The April 20, 2026 Sanriku earthquake had a USGS magnitude of Mw 7.4 (JMA preliminary 7.7, later downgraded to 7.5) — it did NOT reach M8.0, and no M8.0+ event in Japan has been reported since the market opened. Yes
10 STRONG 88 ↓ DOWN article_search The July 30, 2025 M8.8 earthquake struck off Russia's Kamchatka Peninsula — NOT Japan or Japanese territorial waters — and sent tsunami waves to Japan that were downgraded to advisories. This does not satisfy the resolution criteria. Yes
11 MODERATE 72 ↓ DOWN article_search No M8.0+ earthquake has been reported with an epicenter in Japan or Japanese territorial waters in the articles spanning May 2025 to May 2026; all major Japan events were below M8.0. Yes
12 MODERATE 60 ↑ UP kalshi_data The Kalshi market for this question is priced at 58%, with a 30-day increase of +3%, suggesting market participants believe this is more likely than not; this is notably above the naive Poisson base rate of ~31%. Yes
Information Gaps
  • No precise, verified count of M8.0+ earthquakes specifically in Japan or its territorial waters (vs. broader region) from USGS/JMA catalogues was retrieved — the code execution used an assumed count of 10 that may not be precisely verified.
  • No current GPS/geodetic strain accumulation data for the Nankai Trough or other Japanese fault systems was retrieved, which would directly inform near-term rupture risk.
  • No data on slow-slip events or tremor frequency along the Nankai Trough or Japan Trench in 2025-2026, which are key precursory indicators.
  • The exact market open date for KXEARTHQUAKEJAPAN-30 is unclear, making it hard to determine precisely how much of the question window has already elapsed.
  • No information on whether the Kalshi market definition includes only JMA-measured magnitudes or USGS magnitudes, which could differ by 0.2-0.3 units (as seen in the April 2026 Sanriku event).
  • No expert seismological assessment of whether the December 2025 and April 2026 M7.5/7.7 events have released or increased stress on nearby fault segments.
Key Uncertainties
  • The Nankai Trough could rupture within the question window (before 2030) — if it does, it would almost certainly be M8+, but the 30-year probability implies only ~10-15% chance per 4-year window.
  • Ambiguity in magnitude scales (JMA vs USGS) could affect resolution — the April 2026 event was JMA 7.7 but USGS 7.4; a borderline M8.0 event could be disputed.
  • Whether recent M7.5+ earthquakes have increased or decreased stress on nearby fault segments is scientifically uncertain and could affect near-term risk in either direction.
  • The Kalshi price of 58% significantly exceeds the naive Poisson estimate of ~31%, possibly reflecting the Nankai Trough elevated risk, broader territorial waters scope, or market mispricing/thin liquidity.
  • Japan's 'territorial waters' definition could include the Ryukyu Trench, Bonin Islands, and other areas, potentially expanding the relevant seismic zone beyond the main islands.
  • The remaining time window (~3.6 years from May 2026 to Jan 2030) is now shorter than the full question window, reducing remaining probability versus initial estimates.
Market Context
Price: 0.58 · 7d: +0.00 · 30d: +0.03
Volume: LOW-MODERATE · Depth: UNKNOWN
Reliability: LOW-MEDIUM — 118 average daily contracts is modest volume; the wide historical price range (28%-97%) suggests high volatility and possible thin liquidity at extremes; the 58% price significantly exceeds the naive Poisson base rate (~31%), possibly reflecting Nankai Trough risk premium or speculative positioning
4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price
Independent estimate: 34% (45% confidence)
base rate
36%
40% conf
evidence driven
33%
50% conf
contrarian
32%
45% conf
Spread
4pp
Low disagreement
sq1: Is Japan's historical base rate of M8.0+ earthquakes consistent with at least one occurring in any given ~4-year window?
30%
Base rate: ~10 M8+ events in Japan over 126 years gives λ≈0.079/year. Over the remaining ~3.63 years (May 2026 to Jan 2030), Poisson P(N≥1) = 1-exp(-0.079*3.63) ≈ 25%. If we use a slightly higher post-1900 rate accounting for any undercount (λ≈0.10/year), it's ~30%. Japan accounting for 20% of global M6+ events supports the high seismicity. Final base rate estimate: ~30%.
sq2: Are current seismic indicators elevated above baseline?
42%
Nankai Trough 30-year probability of 60-90% for M8-9 implies roughly 10-18% probability per 3.6-year window from Nankai alone. Adding Japan Trench, Sagami Trough, Ryukyu, and Kuril zones, the source-zone-summed probability is plausibly 35-45%. Recent M7.5 (Dec 2025) and M7.7 (April 2026) events triggered megaquake advisories (~1% chance of M8+ in following week, 10x baseline), indicating officials view stress as elevated. However, these events also partially released stress on some segments. Net: elevated above pure base rate, ~40-45%.
sq3: Has a M8.0+ earthquake already occurred since market open?
2%
Confirmed NO: April 2026 Sanriku event was M7.4-7.7 (below 8.0); Kamchatka M8.8 was not in Japanese waters. Small residual probability for definitional ambiguity (JMA vs USGS magnitudes for borderline events). Essentially 2%.
Show full reasoning trace
base rate: Japan has experienced ~10 M8.0+ earthquakes since 1900 (rate ~0.079/year). Over the remaining 3.63-year window, Poisson gives ~25-30% probability. Some upward adjustment for possible undercount and broad territorial waters definition.
evidence updates: Nankai Trough 60-90%/30-year assessment adds ~10-15% via source-specific accumulation. Recent M7.5-7.7 events and megaquake advisories indicate elevated official concern but did not trigger M8+. Source-zone summation across Nankai, Japan Trench, Sagami, Kuril, Ryukyu plausibly elevates prospective probability to ~35-40%. Sq3 confirms no event has occurred yet, slightly reducing remaining probability vs. full window.
combination method: Sq3 acts as a screening factor (essentially 'has it happened already' = No), so we compute forward probability from sq1/sq2 and add small residual. Forward P ~35%, plus 2% residual = ~36%.
final: Combining Poisson base rate (~30%) with elevated source-specific risk (Nankai + other zones, ~40-42%) and the fact that no M8+ has occurred yet, I estimate ~36% probability. This is below recent Kalshi market pricing (~58%) which I suspect over-weights the Nankai Trough headline figures and recent advisories. My confidence is moderate (0.4) due to uncertainty in the precise historical count, magnitude scale ambiguity, and stress redistribution effects.
ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.36, 'confidence': 0.4}, 'evidence_driven': {'p_yes': 0.33, 'confidence': 0.5}, 'contrarian': {'p_yes': 0.32, 'confidence': 0.45}}, 'spread': 0.04, 'individual_p_yes': {'base_rate': 0.36, 'evidence_driven': 0.33, 'contrarian': 0.32}}
5. Devil's Advocate Sonnet · legacy v2 Market price revealed
Reasoning Flaw Missing Info Asymmetric Evidence
Challenges
  1. Potential double-counting concern in sq2: The Nankai Trough 60-90%/30-year figure is an official hazard estimate that already incorporates the historical base rate for that source zone. Adding 'source-zone-summed probability' across Nankai + Japan Trench + Sagami + Kuril + Ryukyu may sum hazards that collectively constitute most of Japan's seismicity — so this isn't additive to the base rate; it's largely the same signal expressed differently. Yet the forecaster treats sq2 as an upward update on sq1.
  2. Conversely, the forecaster may be UNDER-weighting sq2: if Nankai alone gives ~10-18% for the window, and Japan Trench/Kuril/etc. give meaningful independent contributions, source-summation can legitimately exceed the historical Poisson rate because the 126-year catalog may not capture current stress accumulation states. The forecaster acknowledges this but then anchors the final to ~36%, closer to sq1 than sq2.
  3. The historical count of ~10 M8.0+ events is flagged as unverified in information gaps. JMA's catalog (which often runs higher than USGS for Japanese events, as seen in the April 2026 Sanriku case where JMA initially reported 7.7 vs USGS 7.4) could yield a meaningfully higher base rate. If the Kalshi resolution uses JMA magnitudes, the effective base rate is higher than computed.
  4. The combination method is informal. Sq1 (30%) and sq2 (42%) are not independent — sq2 incorporates sq1's information plus source-specific updates. The forecaster ends at 36%, essentially averaging them, but sq2 should arguably dominate since it's the more informed estimate. A principled Bayesian update would likely land closer to 40-45%.
  5. Megaquake advisories in Dec 2025 and April 2026 indicate Japanese officials see elevated short-term risk. The forecaster notes these but then partially discounts them via 'stress release' reasoning without quantitative basis — stress redistribution from M7.4-7.7 events can also LOAD adjacent segments (e.g., Tohoku 2011 was preceded by foreshocks).
Suggested adjustment: +6pp
Adjust upward modestly (not to market, but to ~40%) because: (1) sq2's more-informed estimate should be weighted more heavily than sq1 in combination, since it incorporates source-specific hazard information; (2) the unverified base-rate count and JMA vs USGS magnitude ambiguity asymmetrically favors higher probability (JMA tends to report higher magnitudes for Japanese events, and resolution likely uses JMA); (3) recent megaquake advisories are quantitative signals officials weight seriously, and the 'stress release' counter-argument is speculative. The adjustment is for reasoning issues, not market conformity — the 58% market still likely overshoots.
Market comparison: Forecaster: 33.7%, Market: 58%. The 24-point divergence is substantial but the market has wide historical range (28-97%) and modest volume, suggesting it may overweight Nankai headlines. However, the forecaster's combination method underweights the more-informed sq2 estimate and the JMA magnitude convention. A defensible forecast is ~40%, leaving meaningful but smaller divergence from market.
6. Aggregator Opus · final calibrated forecast
40% (45% confidence)
My independent Poisson base rate analysis (~30% for the 3.63-year remaining window using ~0.079 M8+/year for Japan) combined with source-specific Nankai/Japan Trench hazard estimates yielded 33.7%. I partially accept the Devil's Advocate critique: the combination method did underweight sq2's more-informed source-specific estimate, and JMA's tendency to report higher magnitudes than USGS for Japanese events (demonstrated by the April 2026 Sanriku M7.7 vs M7.4 discrepancy) asymmetrically favors higher probability if Kalshi resolves on JMA. I reject the implication of moving to market: the 58% price reflects thin liquidity (118 contracts/day) and likely overweights Nankai megaquake advisory headlines, and the historical 126-year catalog doesn't support that level. I deviate 18 points below market because the naive Poisson rate is ~31% and even generous source-summation across Nankai+Japan Trench+Kuril rarely exceeds ~45% for a <4-year window.