Scan: When will a supervolcano next erupt?

← Back to scans

When will a supervolcano next erupt? — Before Jan 1, 2050

KXERUPTSUPER-0-50JAN01 · Climate and Weather · 2026-05-08

Agent

19%

Market Price

-13.4%

Edge

70%

Confidence

Volume: 28,689

Spread: 0.1c

Days to resolution: 8638

Markets in event: 1

Final Rationale

Geological base rates anchor this firmly low: strict VEI≥8 eruptions occur once per 50,000-100,000 years, giving ~0.05% over 24 years, while broader VEI≥7 interpretation gives ~3-4%. I partially accept the Devil's Advocate critique on definitional ambiguity—Kalshi-style resolution may be looser than strict geology, justifying a higher weight (~35%) on the broad definition, which lifts my estimate from 4.3% to ~6%. I reject deference to the 19.5% market because it is low-volume, has swung 5-88% historically, dropped 10.5pp in 7 days without geological trigger, and reflects retail overestimation of tail risks rather than informed pricing. The remaining 13-point gap to the market is genuine mispricing, not signal I should chase.

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:

Date	Predicted	Market Price	Confidence
2026-04-30	4%	25%	70%
2026-04-04	4%	32%	50%

1. Planner Opus · Fermi decomposition + tool plan

Tools chosen: wikipedia_lookup article_search web_search kalshi_data code_execution

Sub-questions (Fermi decomposition)

Is the annual base rate probability of a supervolcanic eruption (VEI>=8) greater than 0.01% per year based on geological history? (w=0.35)
Is there currently elevated unrest or scientific warnings at any major supervolcano system (Yellowstone, Toba, Campi Flegrei, Long Valley) that materially increases near-term eruption probability above the geological base rate? (w=0.40)
Does the cumulative probability of at least one supervolcanic eruption over the roughly 24-year window to Jan 2050 exceed 1% when accounting for all known supervolcano systems worldwide? (w=0.25)

Planner reasoning

This question requires estimating the probability of a rare geological event within a ~23.7-year window. The core analytical approach: (1) Establish geological base rates - VEI>=8 eruptions occur roughly 1 per 50,000-100,000 years globally, yielding ~0.05% cumulative probability over 24 years; VEI>=7 events occur ~1 per 700 years, giving ~3-4% over 24 years. The definition of 'supervolcano' is crucial - strict VEI>=8 yields tiny probability, while broader definitions increase it. (2) Check current monitoring signals at major systems: Yellowstone (last erupted ~640,000 BP), Campi Flegrei (has shown recent unrest), Long Valley, Aira caldera. Elevated unrest could shift probability upward from base rate. (3) The temporal analysis: for YES to occur, a specific supervolcano system would need to escalate from current quiescence/mild unrest to full caldera-forming eruption within 24 years. Even systems showing unrest (like Campi Flegrei) have a very long pathway to VEI>=8 eruption. The reference class is clear: no VEI>=8 eruption has occurred in recorded human history. The key uncertainty is whether the question resolves on VEI>=7 or strict VEI>=8, and whether current unrest at monitored systems represents meaningful deviation from background probability. Market prices on Kalshi will provide crowd wisdom as a calibration anchor.

2. Research Parallel tool calls · legacy v2

Tool	Status	Time	Items
wikipedia_lookup	OK	1.5s	-
article_search	OK	0.5s	-
web_search	OK	67.5s	-
kalshi_data	OK	1.3s	-
code_execution	OK	0.0s	-

3. Evidence Synthesis Sonnet · 13 items · legacy v2

#	Strength	Credibility	Direction	Source	Claim	Priced In
1	STRONG	85	↓ DOWN	wikipedia_lookup	A supervolcano is strictly defined as a volcano with a VEI-8 eruption (>1,000 km³ deposits); the most recent VEI-8 eruption was Taupō ~25,600 years ago, and Campi Flegrei/Phlegraean Fields' largest eruptions are only VEI-7, meaning it is not technically a supervolcano.	Yes
2	STRONG	90	↓ DOWN	web_search	USGS estimates Yellowstone's annual probability of a caldera-forming eruption at ~1 in 730,000 (0.00014%); global VEI-8 events occur roughly once every 50,000–100,000 years, giving an annual base rate of 0.001–0.002%, which is well above 0.01% per year only if using broader (VEI>=7) definitions.	Yes
3	STRONG	80	NEUTRAL	code_execution	Computed annual probability for VEI>=8 globally is ~0.002% (1 per 50,000 years), which is below the sq1 threshold of 0.01% per year; VEI>=7 annual probability is ~0.14%, which exceeds 0.01%.	Yes
4	STRONG	88	↓ DOWN	web_search	Yellowstone's magma chamber is only 5–15% molten rock (crystal mush), and there are no signs of impending volcanic eruption based on monitoring data; USGS states the 0.001% annual probability 'is probably an overestimate for the short term.'	Yes
5	WEAK	72	NEUTRAL	article_search	Echinus Geyser at Yellowstone resumed erupting in March 2026 after five years of silence, but this is a geyser (hydrothermal feature), not a sign of magmatic unrest or elevated eruption risk.	No
6	MODERATE	65	↓ DOWN	article_search	No news articles in the past year report elevated volcanic unrest or scientific warnings at Yellowstone, Toba, Campi Flegrei, or Long Valley that scientists say materially increases near-term eruption probability; news about these sites is routine or tourist-related.	Yes
7	MODERATE	82	↓ DOWN	wikipedia_lookup	Phlegraean Fields (Campi Flegrei) is monitored by Vesuvius Observatory and has shown periodic unrest (bradyseism), but its largest known eruptions are only VEI-7, below the strict supervolcano threshold of VEI-8.	Yes
8	MODERATE	75	↓ DOWN	article_search	Recent volcanic activity in the news includes Kilauea (Hawaii, low VEI), Mt. Etna (Italy), Krasheninnikov (Russia, first eruption in centuries after M8.8 earthquake), and Mt. Lewotobi (Indonesia) — none are supervolcano systems and none represent VEI>=7 events.	Yes
9	STRONG	88	↓ DOWN	web_search	USGS monitoring of Yellowstone shows no signs of impending eruption; the magmatic system is described as mostly solid and quiescent as of available monitoring data.	Yes
10	STRONG	80	↓ DOWN	code_execution	Cumulative probability of at least one VEI>=8 eruption globally over 23.65 years is ~0.047%, far below the 1% threshold; even using a broader 10,000-year recurrence interval, the probability is only ~0.24%.	Yes
11	STRONG	85	↓ DOWN	web_search	Scientific literature estimates global VEI-8 recurrence at 5,200 to 48,000 years per event; even using the most optimistic (shortest) estimate of 5,200 years, the 24-year cumulative probability would be ~0.46%, still below 1%.	Yes
12	MODERATE	75	↑ UP	code_execution	If VEI>=7 events (recurrence ~700 years) are included under a broader definition of 'supervolcano,' the 24-year cumulative probability rises to ~3.3%, which exceeds 1%.	Yes
13	MODERATE	45	↑ UP	kalshi_data	Kalshi prediction market prices this event at 19.5% as of current data, a significant drop of 10.5% over the past 7 days, with a wide historical range (5–88%) suggesting high uncertainty and potentially unsophisticated market participants.	No

Information Gaps

No current detailed monitoring data for Campi Flegrei (Phlegraean Fields) unrest levels — recent bradyseism activity and Italian civil protection alert levels were not captured in search results.
No specific data on Toba, Long Valley, or Aira caldera current monitoring status or unrest indicators.
The question's resolution criteria are ambiguous: it is unclear whether the market resolves on strict VEI>=8 or a broader 'supervolcano' definition that could include VEI>=7 events — this definitional issue is the single largest uncertainty.
No information on why the Kalshi market price (19.5%) is so dramatically higher than the geological base rate (~0.05–3.3% depending on definition) — this gap suggests either market mispricing, a broader resolution definition, or systematic overestimation by retail participants.
No data on newly discovered or poorly characterized supervolcano systems that might be omitted from standard global counts.
No peer-reviewed literature on the aggregate global probability across all known supervolcano systems (there are ~20+ known systems worldwide).

Key Uncertainties

Resolution definition: Whether 'supervolcano' in this market means strict VEI>=8 (making probability ~0.05%) or a broader definition including VEI>=7 (raising it to ~3.3%) is the dominant uncertainty.
Kalshi market price at 19.5% is orders of magnitude above the strict geological base rate, raising the question of whether the market is mispriced or incorporates a very broad resolution definition.
Campi Flegrei unrest: This system has shown periodic bradyseism and elevated seismicity in recent years; while its maximum historical eruptions are VEI-7 (not VEI-8), it represents the system most likely to produce a large eruption in the near term, but still far below VEI-8 threshold.
Model uncertainty in geological base rates: The range of 5,200 to 48,000 years between VEI-8 events spans nearly an order of magnitude, creating substantial uncertainty in any probability estimate.
Unknown unknowns: Supervolcano systems that are unmonitored or poorly characterized, particularly in remote regions, could harbor eruption precursors undetected.

Market Context

Price: 0.195 · 7d: -0.105 · 30d: +0.055
Volume: LOW-MODERATE — 153 contracts/day average is thin for a forecasting market · Depth: UNKNOWN — not provided
Reliability: LOW — The 19.5% price is dramatically inconsistent with geological base rates (0.05–3.3%), the wide historical price range (5–88%) indicates high volatility and likely unsophisticated participants, low volume undermines price discovery quality. The sharp 7-day drop of 10.5 percentage points without any geological trigger suggests noise trading rather than information.

4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price

Independent estimate: 4% (68% confidence)

base rate

75% conf

evidence driven

70% conf

contrarian

60% conf

Spread

1pp

Low disagreement

sq1: Is the annual base rate probability of a supervolcanic eruption (VEI>=8) greater than 0.01% per year based on geological history?

Strict VEI>=8 base rate is ~0.001-0.002% per year (1 per 50,000-100,000 years), well below the 0.01% threshold. Only under a broad VEI>=7 definition does it exceed. Most geological literature uses strict definition. Estimate: 8%.

sq2: Is there currently elevated unrest or scientific warnings at any major supervolcano system that materially increases near-term eruption probability above the geological base rate?

15%

Campi Flegrei has shown bradyseism and is monitored, but it is technically VEI-7 max. No scientific warnings of imminent VEI-8 anywhere. USGS describes Yellowstone as quiescent. 'Materially increases' is a high bar—mild unrest doesn't qualify. Estimate: 15%.

sq3: Does the cumulative probability of at least one supervolcanic eruption over ~24 years exceed 1%?

10%

Strict VEI>=8 cumulative is ~0.05-0.46%, below 1%. Only if including VEI>=7 broader definition does it exceed. Scientific consensus uses strict definition. Estimate: 10%.

Show full reasoning trace

base rate: VEI>=8 eruptions occur roughly once per 50,000-100,000 years globally. Over 24 years, cumulative probability is ~0.05%. Even using broadest VEI>=7 definition (recurrence ~700 years), 24-year probability is ~3.3%. No VEI>=8 in recorded human history.

evidence updates: Strong evidence (USGS, geological literature) anchors strict VEI>=8 probability at <0.5% over 24 years. Campi Flegrei unrest pushes up modestly but it's a VEI-7 system. No major supervolcano shows imminent eruption signals. Resolution ambiguity is the main reason to consider higher estimates.

combination method: Sub-questions are threshold indicators about the underlying probability, not direct probabilities of eruption. The weighted average of indicator sub-questions (~11%) overstates the actual eruption probability. The actual question is whether a supervolcano erupts before 2050, which under strict definition is ~0.05% and under broad VEI>=7 is ~3-4%.

final: Weighting strict definition heavily (~75%) with broad definition (~25%) for resolution ambiguity: 0.75*0.005 + 0.25*0.04 = ~0.014. Adding small upward adjustment for unknown systems, monitoring gaps, and Campi Flegrei tail risk gives ~4%.

ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.04, 'confidence': 0.75}, 'evidence_driven': {'p_yes': 0.04, 'confidence': 0.7}, 'contrarian': {'p_yes': 0.05, 'confidence': 0.6}}, 'spread': 0.01, 'individual_p_yes': {'base_rate': 0.04, 'evidence_driven': 0.04, 'contrarian': 0.05}}

5. Devil's Advocate Sonnet · legacy v2 Market price revealed

Missing Info Reasoning Flaw

Challenges

The forecaster identifies definitional ambiguity as 'the single largest uncertainty' but only weights the broad VEI>=7 definition at 25%. Given that Kalshi resolution criteria are typically not strictly geological and retail/market participants may treat large eruptions as qualifying, the broad-definition weight could reasonably be 35-50%. This single parameter swing changes the forecast meaningfully.
Sub-question 3 asks whether cumulative probability exceeds 1% — under the broader VEI>=7 definition the forecaster's own estimate is 3.3%, so the answer should be near-certain YES under that branch. The 10% answer reflects definitional weighting but is then double-counted when the forecaster again applies definitional weighting in the final combination.
The forecaster dismisses the market at 19.5% as 'noise trading' but doesn't seriously engage with the possibility that the market reflects a broader resolution interpretation. If Kalshi/Polymarket-style resolution would count a VEI-7 (e.g., a Campi Flegrei major eruption, or a Tambora-class event), the market price is more defensible than the geological base rate suggests.
Information gap on Campi Flegrei is significant — the forecaster notes recent bradyseism but lacks current alert level data. Italian civil protection raised concerns in 2023-2024; without current data, uncertainty should be wider, not narrower.
The 24-year horizon is long enough that low-probability tail events (unknown systems, monitoring gaps, model error in recurrence intervals) deserve more weight than the small upward adjustment given. Recurrence interval estimates have wide confidence intervals (5,200-48,000 years cited).

Suggested adjustment: +3pp

Modest upward adjustment (from 4.3% to ~7%) is warranted because: (1) the forecaster's own analysis shows VEI>=7 broad-definition gives ~3-4% and definitional ambiguity is acknowledged as the largest uncertainty, justifying higher weight on the broad branch (35-40% rather than 25%); (2) Campi Flegrei monitoring data gaps should widen uncertainty upward; (3) model uncertainty in recurrence intervals is asymmetric (a wrong low estimate matters more than a wrong high one given the floor near zero). This is not adjustment toward the market — it's correcting underweighting of the definitional ambiguity the forecaster identified.

Market comparison: Forecaster: 4.3%, Market: 19.5%. The divergence is large, but the market is low-volume, volatile (5-88% historical range), and likely reflects retail overestimation of supervolcano risk. The forecaster's geological reasoning is sound. However, part of the gap may reflect a broader resolution definition that the forecaster acknowledges but may underweight. A modest upward adjustment to ~7% is justified on reasoning grounds, not market conformity; the remaining gap to 19.5% likely reflects genuine market mispricing.

6. Aggregator Opus · final calibrated forecast

6% (70% confidence)