Scan: Will there be between 11 and 13 earthquakes of magnitude 7.0 or higher worldwide in 2026?

date	predicted	market price	confidence
2026-08-18	12%	10%	68%
2026-08-11	26%	18%	38%

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: polymarket_direct polymarket_related kalshi_related wikipedia claude_news article_search gdelt_news code_execution

Sub-questions (Fermi decomposition)

How many earthquakes of magnitude 7.0 or higher has USGS recorded so far in 2026, and what is today's date relative to year end?
What is the historical annual distribution (mean, variance, min, max) of M7.0+ earthquakes per year in the USGS catalog from ~1990-2025?
Given the 2026 year-to-date count and remaining fraction of the year, what is the Poisson (or overdispersed) probability that the final 2026 total lands between 11 and 13 inclusive?
Are there any recent M6.8-7.1 events whose USGS magnitude may be revised across the 7.0 threshold, or any discrepancy between USGS's significant-earthquakes page and the full catalog that affects counting?
What are the current Polymarket prices for this bucket and the sibling buckets (e.g., 8-10, 14-16, 17+) and do they sum coherently?
Is there currently an elevated seismic sequence (major aftershock zone or subduction rupture) that increases the near-term rate of M7+ events?

Planner reasoning

This is a count-based question best modeled via historical base rates (USGS records ~15 M7+ quakes/year on average, ranging roughly 6-24) conditioned on the year-to-date count in 2026. The key uncertainties are how many M7+ events have already occurred, how much of the year remains, and whether the market price for the 11-13 bucket is consistent with a Poisson/negative-binomial model. Cross-market sibling buckets and news on recent large quakes will refine the anchor.

2. Research Parallel tool calls

tool	status	time	items	summary
polymarket_direct	OK	0.7s	1	## This Market's Polymarket Data Will there be between 11 and 13 earthquakes of magnitude 7.0 or higher worldwide in 2026? - Current price (probability): 14.00% - 7-day price change: +1.00% - 30-day price change: +4.50% - Total volume: $417,329 (USD notional) - Price range: 8.00% - 34.00% - Dat
polymarket_related	OK	5.2s	1	Scanned 100 active Polymarket markets, kept 1 matches. keyword 'earthquakes of magnitude 7.0': 1 markets \| keyword 'earthquake 2026': 0 markets \| keyword 'magnitude 7': 0 markets
kalshi_related	OK	4.9s	2	2 related markets / summaries. keyword 'earthquake': ok \| keyword 'magnitude 7': ok
wikipedia	OK	0.3s	3	Fetched 3 Wikipedia entries (0 missing pages).
claude_news	OK	22.6s	13	Based on my research, here are the key findings: - Current M7.0+ count for 2026 (per USGS/Wikipedia): As of mid-September 2026, Wikipedia's "List of earthquakes in 2026" tracker shows 11 earthquakes in the 7.0–7.9 magnitude range for the year, with none reaching magnitude 8.0 or above . This f
article_search	OK	0.8s	17	Fetched 17 unique articles across 3 queries (lookback=60d). 'magnitude 7 earthquake 2026': 10 articles \| 'USGS major earthquake strikes': 10 articles \| 'earthquake tsunami warning 2026': 10 articles
gdelt_news	OK	66.6s	20	GDELT: 20 articles across 2 queries (lookback=45d). 'magnitude 7 earthquake': 10 hits \| 'strong earthquake USGS': 10 hits
code_execution	OK	60.1s	0	## Key Findings - Historical base rates (1990–2025, n=36 years): Mean = 15.03 M7.0+ quakes/year, sample variance = 13.80, dispersion index (var/mean) = 0.92 — the series is actually slightly under-dispersed relative to Poisson, not over-dispersed. A standard negative-binomial moment-fit i

3. Evidence Brief Sonnet 5 · 6095 chars

# Current state As of mid-September 2026 (~70% of the year elapsed), the Wikipedia/USGS-tracked tally of M7.0+ earthquakes for 2026 stands at **11**, with none reaching M8.0+ (Al Jazeera, Wikipedia "List of earthquakes in 2026," via claude_news). That YTD figure already sits at the low end of the 11–13 target band; the outcome hinges on how many additional M7+ events occur in the remaining ~3.5 months (resolution allows a grace period into Jan 2027 for late-appearing USGS data). # Timeline of key events - 2026-04: Japan Sanriku M7.4–7.5 (JMA 7.7) earthquake, large foreshock/aftershock sequence, no aftershock reached M7 (confirmed, Wikipedia). - 2026-04: Indonesia North Maluku M7.4 (USGS) / M7.3 (BMKG) — agency magnitude discrepancy near threshold (confirmed). - 2026-06: Philippines Mindanao M7.8, >100 deaths (confirmed, Al Jazeera/Wikipedia). - 2026-06: Venezuela M7.2 and M7.5, combined >6,300 deaths — deadliest events of the year (confirmed). - 2026-07: Mexico M7+ event referenced (reported, Al Jazeera). - 2026-07-28: Japan Kyushu/Kumamoto M7.1, deadly, tsunami advisory issued (confirmed, multiple outlets). - 2026-08-14/15: Indonesia Flores M7.7, 47+ deaths, tsunami warning lifted (confirmed, NPR/CNN/CBS). - 2026-08-17: Al Jazeera reports YTD count of "11 above magnitude 7" as of mid-August 2026 (reported). - 2026-09 (date unclear): Colombia earthquake, GCMT Mw7.5 (reported, Wikipedia). - 2026-09 (mid-month): Wikipedia tracker still shows 11 total M7.0–7.9 events for the year, none ≥M8 (reported, claude_news synthesis) — implying no new M7+ event between mid-Aug and mid-Sept. - 2026-09-11: GDELT reports "dozen+ earthquakes in 24hrs" along Ring of Fire, but these are sub-M7 (reported, not counted). # Event Will there be between 11 and 13 (inclusive) M7.0+ earthquakes worldwide in 2026, per USGS? # Outcomes to forecast Yes / No (this bucket vs. all other count-ranges) # Kalshi market anchor Kalshi-direct price not separately returned in tool output; sibling data comes via Polymarket-mirrored ticker. **Polymarket price for this exact 11–13 bucket: 14.00% YES**, up +1% (7d) and +4.5% (30d), range 8–34% over 90 days, volume $417K — treat as the consensus anchor to beat. # Sub-question answers 1. **YTD count & date**: ~11 M7.0+ quakes recorded as of mid-September 2026 (Wikipedia/Al Jazeera), with ~3.5 months (≈30% of year) remaining until close. 2. **Historical distribution (1990–2025)**: Mean = 15.03/yr, variance = 13.80 (dispersion index 0.92, i.e., slightly under-dispersed vs Poisson) [code_execution]. Empirical frequency of years landing in 11–13 = 10/36 = 27.8%. 3. **Poisson probability given YTD=11, ~30% of year left**: remaining-year λ≈4.5; P(final total ∈{11,12,13}) = P(0,1,2 additional) ≈ 17.5% [derived from code_execution model + Poisson calc]. 4. **Magnitude revision risk**: Yes — 2026 already shows agency discrepancies near threshold (North Maluku: BMKG 7.3 vs USGS 7.4; Sanriku: JMA 7.7 vs Mw 7.4–7.5), meaning the final USGS count could shift by ±1 depending on revisions [Wikipedia/claude_news]. 5. **Sibling Polymarket buckets**: 17–19 bucket priced at 31% YES (69% No), volume $292K — notably higher than 11–13's 14%, implying market leans toward a higher final total (consistent with YTD pace toward ~15+ for the year). 6. **Elevated seismic activity**: Yes — active aftershock sequences post-Sanriku (Japan) and North Maluku (Indonesia), plus a cluster of major quakes June–August 2026 (Philippines, Venezuela, Japan, Indonesia), indicate above-average M7+ activity this year, though no single sequence is currently producing M7+ aftershocks. # Key facts (high-confidence, factual) 1. [Wikipedia/Al Jazeera] YTD 2026 M7.0+ count ≈11 as of mid-Sept 2026; none ≥M8. 2. [code_execution] 1990–2025 mean = 15.03/yr; 27.8% of years fall in 11–13 range historically. 3. [Polymarket] 11–13 bucket priced 14%; 17–19 bucket priced 31%. 4. [Wikipedia] Recent years' M7.0-7.9 counts: 2016–2026 ≈ 18,16,6,16,9,9,16,11,19,15,15,(11 partial). 5. [Wikipedia/CNN/NPR] Multiple confirmed M7+ events already in 2026: Sanriku (Apr), North Maluku (Apr), Philippines (Jun, M7.8), Venezuela (Jun, M7.2/7.5), Japan Kumamoto (Jul28, M7.1), Indonesia Flores (Aug15, M7.7), Colombia (Sep, M7.5). # Cross-market signals - Kalshi related: California/Japan M8+ long-horizon markets (18–38% range) — not directly comparable timeframe. - Polymarket: 11–13 bucket = 14% YES; sibling 17–19 bucket = 31% YES — market weight skews toward higher-count buckets, implying an expected final total plausibly in the mid-to-high teens. - No sportsbook equivalent exists for this niche event. # Analyst opinions and speculation - claude_news notes YTD pace (11 by mid-Sept) is "plausible" for landing in 11-13 but flags 3.5 months of remaining risk. - Model-based (code_execution) probability grid shows conditional P(11–13) is highly path-dependent; with YTD near/at the historical mean pace, probabilities in the 20–50% range are possible depending on exact remaining-fraction assumptions. # Directional lean per outcome - **Yes (11–13)**: Supported by current YTD=11 already qualifying at floor of range and historical 27.8% base rate; opposed by ~30% of the year remaining with average pace implying ~4–5 more events, likely pushing total to 14–17 (favoring higher buckets, matching Polymarket's own 17–19 pricing at 31%). - **No**: Favored by cross-market pricing (86% implied by Polymarket) and expected-value math (Poisson remaining λ≈4.5 pushes central estimate above 13). # Gaps / unknowns - Exact resolution methodology: USGS's curated "significant earthquakes" page may not perfectly match the M7.0+ global catalog count used by Wikipedia/Al Jazeera — possible mismatch source. - No direct Kalshi YES price returned (only Polymarket mirror used as anchor). - Precise current-day date/YTD count as of Kalshi close is uncertain beyond mid-September data. # Calibration anchors - Polymarket YES price (anchor): 14% - Poisson-derived estimate given YTD=11, ~30% year remaining: ~17-18% - Historical empirical base rate (any year, no conditioning): ~24-28%

4. Forecasters Fable 5.1 + GPT-5.6 Sol · in parallel

claude-fable-5-1

58% conf

Yes	17%
No	83%

YTD count is 11 as of mid-September, already at the floor of the 11–13 band, but ~30% of the year remains and the 1990–2025 average rate (15.0/yr) implies λ≈4.4 additional M7+ events; a Poisson calculation gives P(0–2 more) ≈ 18%, with slight under-dispersion not materially changing this. The market anchor (Polymarket 11–13 bucket at 14%, sibling 17–19 at 31%) agrees that the central expectation for the final total is ~15–16, above the band. Magnitude revision risk (North Maluku USGS 7.4 vs BMKG 7.3) is a modest two-sided wildcard that could shift the count by ±1. I land slightly above the market anchor given the model estimate, at 17% Yes.

gpt-5.6-sol

68% conf

Yes	16%
No	84%

The tally of 11 by mid-September means Yes requires no more than two additional M7.0+ earthquakes during the remaining roughly 30% of 2026. Historical rates imply about 4.5 additional events, and the conditional Poisson estimate gives the 11–13 bucket a 17.5% chance, while recent elevated activity and the sibling 17–19 bucket's 31% price point toward a final count above 13. The 27.8% unconditional historical frequency for 11–13 is less informative after conditioning on the current tally, while magnitude revisions create modest boundary uncertainty. Anchoring on the 14% market price, I move slightly upward to 16% because the count model directly conditioned on 11 supports a somewhat higher probability.

5. Devil's Advocate Sonnet 5 · red-team critique

Both forecasts converge almost mechanically on the ~17-18% Poisson estimate, but neither interrogates the model's key assumption that remaining-year λ≈4.5 applies uniformly — the brief notes a "cluster of major quakes June–August" and "elevated seismic activity" with active aftershock sequences (Sanriku, North Maluku), yet zero new M7+ events were recorded between mid-Aug and mid-Sept, suggesting the recent pace has actually slowed rather than continuing at the historical average rate; this stall could materially raise P(Yes) but neither forecaster adjusts for it.

Neither forecaster explicitly reconciles the tension between the empirical base rate (27.8%) and the Poisson-conditional estimate (17.5%) beyond noting the latter is "more informative" — but with only ~3.5 months left and a small integer count, the Poisson model's discreteness means small errors in λ swing the answer a lot, and no sensitivity analysis (e.g., what if λ=3.5 instead of 4.5, given recent lull) is shown.

The magnitude revision risk is flagged by both as "modest" and roughly symmetric, but the evidence lists three separate near-threshold discrepancies in 2026 alone (North Maluku, Sanriku, plus the Colombia M7.5 GCMT vs. other agency figures) — this unusually high rate of borderline classifications this specific year suggests boundary risk is elevated beyond a typical year and could push the actual USGS-resolved count either up or down from the Wikipedia tally of 11, which neither forecast quantifies as a discrete scenario.

Both anchor tightly to the Polymarket 14% price and converge to 16-17%, but neither questions whether Polymarket's $417K volume and the 8-34% 90-day range (a 26-point swing) indicates a thin, volatile market that may not have fully priced in the most recent lull in events (mid-Aug to mid-Sept no new quakes) — the 30-day uptrend (+4.5%) actually argues for some upward momentum in the Yes price that both forecasts underweight when landing just above the anchor.

Neither forecast considers the possibility that the resolution source mismatch (USGS "significant earthquakes" list vs. Wikipedia's broader catalog, explicitly flagged as a gap) could produce a lower official count than the Wikipedia-tracked 11, which is a scenario favoring Yes (or even a lower bucket) that isn't reflected in either probability.

Both forecasters use nearly identical language and land within 1 point of each other (17% vs 16%), suggesting anchoring/herding on the same Poisson figure rather than independent reasoning — this convergence itself is suspicious given the wide historical dispersion (2016-2025 counts ranging from 6 to 19) that should produce more forecast uncertainty than a tight 16-17% consensus implies.

6. Aggregator Fable 5.1 · final calibrated forecast

17% (63% confidence)

Yes: 17% No: 83%

With ~11 M7.0+ events by mid-September and ~29% of the year remaining, Yes requires no more than two additional M7+ quakes; a Poisson model at the 1990–2025 rate (λ≈4.4) gives ~18%, while using the lower 2016–2025 mean (13.5/yr, λ≈3.9) pushes it toward ~25%, so the critique's sensitivity point is valid but bounded. Offsetting this upward pull, the brief is ambiguous on whether the September Colombia Mw7.5 is already in the tally of 11 (Al Jazeera reported 11 in mid-August), and if the true count is 12 the probability of ≤1 more event drops to ~7%; magnitude-revision risk is roughly symmetric. The one-month lull is not predictive under a Poisson process, so I don't credit it. Netting these, I land slightly above the Polymarket anchor (14%) and in line with both forecasters at 17% Yes.

Pipeline Timing

Total pipeline time: 206.9s

Per-tool research timings shown in the Research section above.