Tools chosen: kalshi_direct kalshi_related polymarket_related wikipedia claude_news gdelt_news code_execution

Sub-questions (Fermi decomposition)

Planner reasoning

This is a pure base-rate seismology question: the probability of at least one M8.0+ earthquake with epicenter in Japan/its territorial waters over roughly a 4-year window (2026-2029). The key research is the historical frequency of M8.0+ events in the Japan region, official JMA/Japanese government probability forecasts (e.g., Nankai Trough 30-year probabilities), and the Kalshi market's current price as the consensus anchor. Poisson math converts the base rate into a window probability.

# Current state No M8.0+ earthquake with an epicenter in Japan or its territorial waters has been confirmed since the market's tracking period began; the last qualifying historical event was the 2011 Tōhoku M9.0-9.1 quake. Kalshi currently prices YES at 40%, down sharply from a 77% high and down 18 points in the last 30 days, even as a major M7.7 Japan Trench quake and a deadly Kumamoto earthquake occurred in 2026 — suggesting the market is discounting recent large-but-sub-8.0 events and/or time decay as the 2030 deadline approaches. # Timeline of key events - 1923-09-01: Great Kantō earthquake, M8.1 (confirmed, historical precedent — not within current market window) - 1933-03-03: Sanriku earthquake, M8.4 (confirmed, historical) - 1944-12-07: Tonankai earthquake, M8.1 (confirmed, historical) - 1946-12-21: Nankaido earthquake, M8.1-8.4 (confirmed, historical) — Nankai Trough's most recent megaquake - 1968-05-16: Tokachi-oki earthquake, M8.2-8.3 (confirmed, historical) - 2003-09-25: Tokachi-oki earthquake, M8.3 (confirmed, historical) - 2011-03-11: Tōhoku earthquake, M9.0-9.1 (confirmed) — largest in Japan since 1900, occurred well before current contract window - 2024-08: M7.1 Hyūga Sea earthquake triggers JMA's first-ever "massive earthquake" advisory for Nankai Trough (reported, Nippon.com) - 2025-09-27: Earthquake Research Committee revises 30-year Nankai Trough M8-9 probability from "~80%" framing to a 60%-94.5%+ range (with an alternative model giving 20-50%) (reported, Japan Times) - 2026-01-16: Panel raises 30-year probability of M7.8-8.5 Chishima Trench (Nemuro, Hokkaido) quake to ~90% (reported, Japan Times) - 2026 (early, precise date unclear): M7.7 Japan Trench earthquake occurs; JMA calculates ~1% short-term chance of M8+ aftershock/follow-on in following days ("10x normal") (reported, Scientific American) — did NOT meet M8.0 threshold itself - 2026-07 (~14-30): Kumamoto-area earthquake with rising death toll draws heavy US/international coverage; magnitude not specified in available research — status as M8+ event is UNCONFIRMED (reported only via headline coverage, gap) # Event Will an M8.0+ earthquake occur with epicenter in Japan or its territorial waters before Jan 1, 2030? # Outcomes to forecast Yes / No # Kalshi market anchor KXEARTHQUAKEJAPAN-30 current YES price: **40%**. Down 3 points over 7 days, down 18 points over 30 days. Price range over its life: 31%-77%. Average daily volume ~182-283 contracts (146-68 data points depending on feed) — moderate liquidity. The steep 30-day decline is notable and unexplained by any confirmed qualifying event; likely reflects time decay plus market reassessment after recent large-but-sub-8.0 quakes. # Sub-question answers 1. **Kalshi YES price/trend** — 40% currently, down from a 77% high; -18% in 30 days, -3% in 7 days; volume ~182-283 contracts/day (kalshi_direct, kalshi_related). 2. **M8.0+ count since 1900** — VolcanoDiscovery cites ~9-10 events M8.0-9.0 plus the 2011 M9.1, implying ~0.07/yr (1 per ~14 years) region-wide; code_execution analysis narrows a defensible catalog count to n≈7-8 strict Japan-epicenter events, giving λ≈0.056-0.064/yr. 3. **Territorial-waters restriction** — Not directly quantified by any tool; many cataloged "Japan region" M8+ events (e.g., 2006 Kuril M8.4, parts of Tōhoku sequence) occur on offshore subduction zones that may or may not fall within Japan's territorial waters proper vs. EEZ/high seas, meaning the strict resolution criterion could yield a lower count than the broad regional catalog. This is a genuine ambiguity/gap. 4. **Official JERC probability estimates** — Nankai Trough: 60%-94.5%+ over 30 years (alt. model: 20-50%) (Japan Times, Sept 2025). Chishima Trench/Nemuro: ~90% over 30 years (Jan 2026). Tokachi (Hokkaido): ~20% over 30 years, ~50% over 50 years. Northern Sanriku (Japan Trench): 4-20% over 30 years. Combined "mega-earthquake" scenario: 7-40% (claude_news synthesis of JERC data). 5. **Magnitude scale / recent M8+ events** — JMA uses Shindo for intensity, not magnitude; Mw/JMA magnitude used for earthquake sizing (Wikipedia). No confirmed M8.0+ Japan-epicenter event has occurred in 2024-2026 per available research; the 2025 Kamchatka M8.8 is explicitly outside Japan's territorial scope. The 2026 M7.7 Japan Trench quake did not meet the threshold. 6. **Poisson probability** — Central estimate ≈20-25% for a ~4-year window (2026-2030) using n≈7-8 historical events since 1900 (code_execution); sensitivity range 15%-32% across plausible event-count assumptions (n=5 to n=12). # Key facts (high-confidence, factual) 1. [kalshi_direct] Current YES price 40%, down 18pts over 30 days. 2. [VolcanoDiscovery via claude_news] ~0.07 M8+ quakes/year historically in Japan region (1-per-14-years). 3. [Japan Times, 2025-09-27] Nankai Trough 30-yr M8-9 probability revised to 60-94.5%+. 4. [Japan Times, 2026-01-16] Nemuro/Chishima Trench 30-yr probability ~90%. 5. [Wikipedia] 2011 Tōhoku M9.0-9.1 is the largest Japan-region quake since 1900; last qualifying M8+ event. 6. [code_execution] Poisson base rate gives ~20-25% central probability for 2026-2030 window. # Cross-market signals - Kalshi related: California M8+ before 2035 priced 38%; California M8+ before 2028 priced only 13% — suggests near-term (shorter window) contracts price much lower, consistent with Poisson time-scaling. - Polymarket: No matching markets found (polymarket_related). - Sportsbook implied: N/A (not applicable to this category). # Analyst opinions and speculation - code_execution flags that if Kalshi prices well above ~30-35%, it likely reflects JERC's high 30-yr Nankai/Japan Trench estimates rather than pure historical Poisson counting; if well below ~15%, market may underweight historical frequency. - Elevated short-term stress-based risk after the 2026 M7.7 Japan Trench quake (JMA: ~1% chance of M8+ "in coming days," 10x baseline) shows event clustering/non-stationarity is a live concern, though this is a transient signal, not a structural rate change. # Directional lean per outcome - **Yes**: Elevated official 30-yr JERC probabilities (60-94.5% Nankai, ~90% Nemuro) imply meaningfully above-baseline annual hazard; Japan is in an active seismic period (2024 Hyūga Sea, 2026 Japan Trench M7.7, 2026 Kumamoto quake) signaling regional stress; Poisson base rate alone gives 20-25%, and JERC-adjusted estimates could push higher. - **No**: Strict "territorial waters" epicenter criterion may exclude many historically-counted regional events, lowering effective rate; historical M8+ recurrence is genuinely infrequent (~14-yr average) and 2011's M9 may have released significant regional strain reducing near-term Nankai risk in some models; Kalshi's own steep 30-day price decline to 40% suggests market participants are revising downward despite recent seismic activity. # Gaps / unknowns - Whether 2026 Kumamoto earthquake reached/approached M8.0 is unconfirmed in research (headline-only coverage, no magnitude cited) — material gap given proximity to threshold discussions. - Exact legal/geographic definition of "territorial waters" vs. EEZ for resolution purposes not confirmed. - No official Kalshi rules text beyond one-line description; no explicit magnitude-scale designation (JMA vs Mw) for resolution. - Reason for Kalshi's 30-day price drop (-18pts) not explained by any confirmed resolving event — could be market mispricing or private information. # Calibration anchors - Kalshi current YES price: 40% (anchor). - Historical Poisson base rate for ~4-yr window: ~20-25% (code_execution). - JERC 30-yr Nankai Trough probability: 60-94.5%; Nemuro 30-yr: ~90% — these are 30-yr windows, roughly 7-8x longer than the ~4-yr market window, so naive linear scaling would suggest much lower short-window probabilities (~10-15% if scaled naively from 60-90%/30yr).

1. Both forecasters converge almost identically (0.30 vs 0.32) via nearly the same blend of Poisson base rate and Kalshi anchor, suggesting herding on one methodology rather than independently stress-testing the wide JERC probability range (60-94.5% Nankai vs. alt-model 20-50%), which could justify a higher or more uncertain estimate. 2. Neither forecast meaningfully engages with the unresolved Kumamoto earthquake gap ("status as M8+ event is UNCONFIRMED... material gap given proximity to threshold discussions") — if this event turns out to be near-M8, it materially changes the near-term hazard picture and possibly the resolution itself, yet both dismiss it in one clause. 3. Both treat the Kalshi 40% anchor as a meaningful data point to average toward, despite the brief explicitly noting "the steep 30-day decline is notable and unexplained by any confirmed qualifying event" and "could be market mispricing or private information" — neither forecaster interrogates whether the anchor deserves less weight given this unexplained volatility and only moderate liquidity (~182-283 contracts/day). 4. The territorial-waters vs. EEZ ambiguity is flagged as "a genuine ambiguity/gap" that could exclude events like the 2006 Kuril M8.4 or parts of the Tōhoku sequence, but both forecasters only qualitatively note it "trims" the rate rather than widening their uncertainty band or considering it could cut the effective historical n significantly below 7-8, pulling probability toward the low end of the 15-32% sensitivity range. 5. Neither forecaster considers post-2011 stress-shadow effects (reduced near-term Nankai risk after the M9.1 released regional strain) as a distinct scenario versus the clustering/non-stationarity argument (active 2024-2026 sequence signaling elevated risk) — these two effects point in opposite directions and neither is explicitly weighed against the other. 6. Both forecasts sit comfortably inside a narrow 30-32% band with no tail-scenario consideration (e.g., a JERC-implied cluster-driven Nankai event actually firing before 2030, which alone carries a nontrivial multi-percent annualized chance per the 60-94.5%/30yr range), understating the right-tail risk that Kalshi's historical 77% high price may have been partially pricing.

30% (56% confidence)

Yes: 30% No: 70%

The remaining window is only ~3.4 years, and a strict Japan-epicenter M8.0+ catalog implies λ≈0.056-0.07/yr, giving a Poisson base rate of roughly 17-25%; the territorial-waters restriction (excluding Kuril/high-seas events) pushes toward the low end of that range rather than above it. JERC's headline 30-year figures (Nankai 60-94.5%, Nemuro ~90%) sound alarming but scale to only ~10-20% over this window, and the alternative Nankai model (20-50%/30yr) widens uncertainty downward as much as upward. Kalshi's 40% deserves partial but discounted weight: liquidity is moderate, the price has fallen 18 points without any resolving event, and comparable contracts (California M8+ by 2028 at 13%) show these long-horizon disaster markets carry a persistent YES premium. The devil's advocate is right that the unconfirmed Kumamoto magnitude and clustering/non-stationarity argue for right-tail respect, but post-2011 stress-shadow effects cut the other way and the 2026 M7.7 and Kumamoto events are confirmed sub-threshold. I settle modestly below the market anchor and just above the pure base rate at 30%.

Total pipeline time: 221.0s