Scan: Will more than 8 tropical cyclones make landfall in China during 2026?

1. Planner Fable 5.1 · Fermi decomposition + tool plan

Tools chosen: polymarket_direct polymarket_related kalshi_related claude_news article_search gdelt_news wikipedia code_execution

Sub-questions (Fermi decomposition)

How many tropical cyclones has the CMA (or Chinese media citing CMA/NMC) recorded as making landfall in China so far in 2026, and what are their names and dates?
What is the historical distribution of annual CMA landfalling tropical cyclone counts (1949–2025), and in what fraction of years did the count exceed 8?
Given the count at the current date, what is the climatological distribution of additional China landfalls from now through December (August–November landfall frequency by month)?
What is the current and forecast ENSO state (La Niña/El Niño/neutral) for late 2026, and what do seasonal outlooks (CMA, HKO, TSR, JTWC/JMA) say about Western North Pacific activity and landfall risk for China this season?
How does the CMA 'Landfalling TCs' page define China landfall — does it include Taiwan, Hainan, Hong Kong, and Macau — and did recent years (2018 with 10, 2024 with ~8-9) show this inflating counts relative to mainland-only tallies?
What do the sibling Polymarket brackets (e.g., 6, 7, 8, 9, 10+ landfalls) imply about the crowd's full distribution, and are they internally consistent?

Planner reasoning

This is a count-threshold question (9+ CMA-recorded landfalls in China in 2026) that decomposes into: the running 2026 tally so far, the climatological distribution of CMA annual landfall counts (which historically average ~7, with 9+ occurring roughly a quarter of years), the ENSO/seasonal forecast for the remaining active months, and definitional quirks in how CMA counts landfalls (e.g., inclusion of Taiwan/Hainan/Hong Kong). The Polymarket bracket prices are the primary anchor, with sibling brackets giving an implied distribution to sanity-check.

2. Research Parallel tool calls

tool	status	time	items	summary
polymarket_direct	OK	0.2s	1	## This Market's Polymarket Data Will more than 8 tropical cyclones make landfall in China during 2026? - Current price (probability): 35.00% - 7-day price change: -42.50% - 30-day price change: +10.50% - Total volume: $20,645 (USD notional) - Price range: 12.50% - 77.50% - Data points: 41 days
polymarket_related	OK	5.2s	0	Scanned 100 active Polymarket markets, kept 0 matches. keyword 'tropical cyclones landfall China': 0 markets \| keyword 'typhoon China 2026': 0 markets \| keyword 'typhoon landfall': 0 markets
kalshi_related	OK	5.0s	2	2 related markets / summaries. keyword 'typhoon': no matches \| keyword 'tropical cyclone China': ok \| keyword 'hurricane season': ok
claude_news	OK	42.2s	13	Key findings on 2026 China typhoon landfalls (as tracked by CMA/NMC) and outlook - Typhoon Maysak (美莎克, CMA No.10) – on July 3, a tropical depression in the South China Sea strengthened into a tropical storm and was designated by the National Meteorological Centre as this year's No.10 typh
article_search	OK	46.8s	30	Fetched 30 unique articles across 4 queries (lookback=120d). 'typhoon makes landfall China 2026': 10 articles \| '台风登陆 2026 中国气象局': 10 articles \| 'China Meteorological Administration typhoon season 2026 outlook': 10 articles \| 'Western Pacific typhoon season 2026 La Nina El Nino forecast': 10 articl
gdelt_news	OK	139.0s	30	GDELT: 30 articles across 3 queries (lookback=90d). 'typhoon landfall China': 15 hits \| 'tropical storm landfall Guangdong Hainan Fujian Zhejiang': error GDELT rate-limited after retries (429) \| 'typhoon Taiwan landfall 2026': 15 hits
wikipedia	OK	0.2s	4	Fetched 4 Wikipedia entries (0 missing pages).
code_execution	OK	44.7s	0	Key Findings — China Landfalling TC Count 2026 (>8 landfalls, i.e., ≥9) - Historical base rate (2015–2025, n=11 yrs): mean = 7.0, variance = 2.4 (var < mean → underdispersed, not overdispersed as often assumed for TC counts). Only 2 of 11 years (18.2%) had >8 landfalls (2018:10, 2024:9)

3. Evidence Brief Sonnet 5 · 7736 chars

# Current state As of the most recent research (early September 2026, ~10 months before contract close), the CMA/NMC has recorded at least 4 confirmed China landfalls (Maysak, Bavi, Noul, Dolphin) with a 5th (Saudel) forecast/reported and a possible 6th (Narra, unconfirmed as landfall vs. "affecting"). CMA's official April 2026 outlook called for 7–9 total landfalls this year, and its September 2 autumn update forecasts 2–4 additional landfalls through year-end — putting the plausible full-year total in the 6–10 range, with ">8" (i.e., ≥9) a genuine toss-up pending final tally and CMA's counting conventions. # Timeline of key events - 2026-04-01: CMA National Climate Center seasonal outlook: 24–26 NW Pacific/SCS storms forecast, 7–9 expected to make China landfall, above-average intensity (confirmed, per Baidu/Sina reporting of CMA statement). - 2026-07-03/04: Tropical Storm Maysak (CMA No.10) makes first landfall of season near Sanya/Qionghai, Hainan, second landfall near Guangxi/Vietnam border (confirmed, people.com.cn). - 2026-07-11: Typhoon Bavi makes landfall at Taizhou, Zhejiang after earlier Rota (N. Mariana) landfall at peak intensity; ~2.8M evacuated in China (confirmed, multiple outlets: CNBC, CBC, CBS). - 2026-07-26: Typhoon Noul ("Hongxia," CMA No.12) makes landfall in Huidong, Guangdong at severe typhoon intensity; called strongest China landfall of year at the time (confirmed, Al Jazeera, Baidu, GDELT cluster). - 2026-08-09: Typhoon Dolphin (CMA No.13) makes landfall near Wenzhou, Zhejiang at severe typhoon intensity; >1M evacuated (confirmed, CNBC, Al Jazeera). - 2026-08-28: Typhoon Saudel forecast/reported to make landfall between Zhejiang and Fujian; Tropical Storm Narra concurrently affecting Guangdong/Guangxi/Hainan (reported, not fully confirmed as completed landfall in sources). - 2026-09-01/09: Typhoon Krovanh forms but does NOT make China landfall — makes landfall in Japan instead (confirmed negative data point, worlddata.info). - 2026-09-02: CMA autumn forecast: 8–10 more NW Pacific/SCS storms expected in autumn (slightly below 10.7 average), 2–4 expected to make landfall/affect China (near/slightly above 2.2 average) (confirmed, Sina/CMA statement). # Event Will more than 8 tropical cyclones (i.e., ≥9) make landfall in China during calendar year 2026, per CMA Best Track Dataset (or ESCAP/WMO fallback)? # Outcomes to forecast - Yes (>8 landfalls, i.e. ≥9) - No (≤8 landfalls) # Kalshi market anchor No kalshi_direct price was returned in this research pull (only kalshi_related keyword scans, which found no matching typhoon/China-landfall markets). **Gap: true Kalshi consensus price is unavailable; use Polymarket sibling market (35% as of latest snapshot, down sharply from a 77.5% high) as the best available cross-market proxy.** # Sub-question answers 1. **CMA landfall count so far in 2026** — At least 4 confirmed CMA-numbered landfalls: Maysak (No.10, July 3, Hainan), Bavi (July 11, Zhejiang), Noul/Hongxia (No.12, July 26, Guangdong), Dolphin (No.13, Aug 9, Zhejiang); Saudel (Aug 28, Zhejiang/Fujian border) reported but not fully confirmed complete; Narra possibly a 6th, unconfirmed. [claude_news, article_search, gdelt_news] 2. **Historical distribution 1949–2025** — 2015–2025 mean ≈7.0/year, variance 2.4 (underdispersed); only 2 of 11 years (18%) exceeded 8 (2018: 10, 2024: 9). [code_execution] 3. **Climatological additional landfalls Aug–Dec** — Aug–Dec typically accounts for ~55–60% of annual landfalls; remaining-season λ≈4.0–4.4 after early August. [code_execution] 4. **ENSO/seasonal outlook** — CMA's official outlook (Apr 2026) calls for above-average landfalls (7–9) and above-average intensity; a Baidu wiki source attributes this to "strong El Niño development" but this is uncorroborated by an official ENSO advisory and should be treated as rumored. [claude_news] 5. **CMA landfall definition scope** — Not explicitly detailed in retrieved research; historical inflation via Taiwan/Hainan/HK/Macau inclusion (10 in 2018, ~9 in 2024) is asserted by the sub-question framing but not directly confirmed by a CMA methodology excerpt in this research pull. **Gap.** 6. **Polymarket sibling brackets** — Not found; polymarket_related scan returned zero matching sibling markets (6/7/8/9/10+ brackets). Only the single ">8" market itself was retrievable. **Gap.** # Key facts (high-confidence, factual) 1. [code_execution] 2015–2025 empirical base rate: P(>8 landfalls) ≈ 18% (2/11 years); Poisson-model estimate ≈27–30%; blended unconditional estimate ≈20–28%, centered ~25%. 2. [claude_news, Sina] CMA's own April 2026 forecast: 7–9 China landfalls for the full year — its central/upper estimate already sits at or above the ">8" threshold. 3. [CNBC, Al Jazeera, GDELT cluster] Four typhoons (Maysak, Bavi, Noul, Dolphin) confirmed China landfalls by Aug 9, 2026 — an above-average pace for that point in the season. 4. [Sina, 2026-09-02] CMA autumn forecast: 2–4 additional China landfalls expected Sept–Nov 2026. 5. [worlddata.info] Krovanh (formed Sept 1) did NOT make China landfall, landing in Japan instead — a reminder that near-misses are common and not all storms count. 6. [Polymarket] Market price fell sharply from a 77.5% high to 35% over the past 7 days (-42.5pp), despite a +10.5pp 30-day rise — recent repricing lower. # Cross-market signals - Kalshi related: no directly relevant sibling markets found (China EUV, GDP, trade markets irrelevant/noise). - Polymarket: 35% YES (current), high volatility (12.5%–77.5% range over 41 days), volume modest ($20.6k). Sharp 7-day drop suggests recent news (e.g., Krovanh missing China, or slower pace after Dolphin) pushed sentiment down. - Sportsbook implied: none available (not applicable to this event type). # Analyst opinions and speculation - claude_news synthesis: "running tally... putting the full-year total plausibly in the 7–10 range, making >8 a close call." - code_execution modeling: conditional on K≈4–5 landfalls by early August (matching actual confirmed count), P(total≥9) ≈ 37–64%, notably higher than the unconditional ~25% base rate — implying current pace favors YES more than pure climatology would suggest. # Directional lean per outcome - **Yes (>8):** Supported by above-average confirmed pace (4 confirmed landfalls by Aug 9, possibly 5–6 by end of August), CMA's own 7–9 outlook (upper end already exceeds threshold), and active autumn forecast (2–4 more). Conditional models put this scenario at 45–65% given the pace so far. - **No (≤8):** Supported by historical base rate (only 18% of recent years exceeded 8), Krovanh's miss showing not every storm makes landfall, underdispersed historical variance (extreme years are rare), and Polymarket's recent sharp repricing down to 35%. # Gaps / unknowns - No Kalshi direct YES price retrieved for this specific ticker — must infer from Polymarket proxy. - Saudel/Narra landfall status not fully confirmed as completed in sourced articles (dated as forecasts/alerts). - No confirmation of exact CMA methodology re: Taiwan/Hainan/HK/Macau inclusion for 2026 counting. - No visibility into Sept–Dec 2026 actual events beyond the Sept 2 forecast (research lookback ends ~mid-Sept). - Sibling Polymarket bracket markets (6,7,9,10+) not found — cannot cross-check crowd's full distribution consistency. # Calibration anchors - Polymarket YES price (proxy anchor): 35% (down from 77.5% high; up from 30-day-ago level by +10.5pp net despite the 7-day plunge). - Historical base rate: ~18% of years (2015–2025) exceeded 8 landfalls. - Poisson/blended unconditional model: ~20–28%, centered ~25%. - Conditional model given ~4–5 confirmed landfalls by early August 2026: ~37–64%, pivot point at K=5 (~60%).

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

claude-fable-5-1

45% conf

Yes	36%
No	64%

Four CMA landfalls are confirmed by Aug 9 (Maysak, Bavi, Noul, Dolphin), with Saudel likely a 5th and Narra an uncertain 6th, putting the count at ~5–6 entering September. CMA's Sept 2 autumn outlook expects only 2–4 more landfalls (near the 2.2 average), so reaching ≥9 requires the high end of both the current tally and the autumn forecast — plausible but less than even odds, and Krovanh's miss shows near-misses are common. The historical base rate (~18% of 2015–2025 years exceeded 8) and underdispersed variance argue for No, while the above-average pace and CMA's 7–9 seasonal forecast keep Yes alive. No Kalshi direct price exists; the Polymarket proxy at 35% (after a sharp drop from 77.5%) aligns closely with my conditional estimate, so I anchor near it.

gpt-5.6-sol

56% conf

Yes	37%
No	63%

The best market anchor is Polymarket at 35% YES, since no direct Kalshi price was available. Four landfalls are confirmed and Saudel may be a fifth, while CMA forecasts 2–4 additional autumn landfalls; reaching nine therefore likely requires the current count to be at least five and the remainder to land near the forecast's upper end. The above-average pace and conditional models support a slight upward adjustment from the market, but CMA's full-year 7–9 outlook, the 18% recent historical base rate, and uncertainty over Saudel and Narra keep NO favored. I therefore place YES at 37%, close to the proxy consensus.

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

Both forecasts converge tightly around the Polymarket proxy (35–37%) despite the brief's own conditional model (given K≈4–5 by early August) putting P(≥9) at 37–64%, "pivot point at K=5 (~60%)" — since Saudel/Narra plausibly bring the count to 5-6 already, both forecasts arguably under-weight this conditional evidence in favor of anchoring on a thin, volatile, low-volume ($20.6k) proxy market from a different platform.

Neither forecast grapples with the ambiguity in sub-question 5 (CMA counting conventions re: Taiwan/Hainan/HK/Macau) — historical inflated years like 2018 (10) and 2024 (9) may reflect broader counting scope, meaning the "≥9" threshold could be easier to hit than the strict mainland-landfall reading both forecasts implicitly assume.

Both treat Saudel as merely "likely" a confirmed landfall and Narra as fully uncertain, effectively defaulting to the low end (4-5 confirmed) rather than presenting a range or scenario tree — this systematically pulls their estimate toward the lower conditional-model outputs (37%) rather than the higher ones (64%) that would apply if Saudel+Narra both count.

Neither forecast flags the Polymarket price's extreme volatility (12.5%–77.5% over 41 days, sharp 7-day drop) as a reliability concern; treating a market that swung 42.5pp in a week as a stable "consensus" anchor is questionable, especially given the brief explicitly warns "true Kalshi consensus price is unavailable."

Both forecasts cite the 18% historical base rate but don't reconcile it with the fact that CMA's own official forecast (7–9) already implies a ~40-50%+ chance of hitting the upper bound in a "normal" outlook year — i.e., the base rate is drawn from unconditional history, while 2026 is explicitly flagged by CMA as an above-average year, making the historical base rate a weaker reference class than the conditional pace-based model.

Neither considers the CMA's own upper-bound outlook (9, the top of the 7-9 range) as a distinct scenario worth separate weight — if CMA's forecast is taken at face value, "exactly 9" sits right at the >8 threshold, meaning the true probability could be meaningfully higher than 35-37% if CMA's forecasting track record is decent, a consideration absent from both rationales.

6. Aggregator Fable 5.1 · final calibrated forecast

38% (50% confidence)

Yes: 38% No: 62%

Four CMA landfalls are confirmed through Aug 9, Saudel is probably a fifth, and Narra is an uncertain sixth, so the tally entering autumn is most likely 5 (possibly 6). CMA's Sept 2 outlook expects 2–4 more storms to landfall or affect China, and Krovanh's Japan landfall already consumed one autumn storm without counting, so reaching ≥9 requires the upper end of both the current count and the autumn forecast — roughly a 25–30% chance if the count is 5 and ~50% if it is 6. The critique correctly notes that both forecasts leaned toward the low end of the conditional model and that CMA's own 7–9 outlook makes the 18% unconditional base rate a weak reference class, which justifies a modest nudge above the thin, volatile Polymarket proxy at 35%. However, the sharp recent repricing lower likely reflects real information (e.g., Saudel/Narra status or a slower September), so I stay near the market rather than moving to the model's optimistic end.

Pipeline Timing

Total pipeline time: 269.8s

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