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 · 2026-04-30

Agent

25%

Market Price

-21.0%

Edge

70%

Confidence

Volume: 27,572

Spread: 5.0c

Days to resolution: 8646

Markets in event: 1

Final Rationale

Geological base rates for VEI≥8 eruptions over a 24-year window yield only 0.003%–0.14%, anchoring my estimate near ~2%. I partially accept the Devil's Advocate critique on resolution ambiguity: 'supervolcano eruption' is contested, and over 23.6 years, modest activity at Campi Flegrei (ongoing bradyseism since 2012), Kikai (magma recharge), or other under-monitored systems could plausibly resolve YES under a loose interpretation, justifying raising my estimate to ~4%. I reject the market's 20% as a reliable signal — thin liquidity (154 contracts/day), shallow orderbook, wide historical range (5–88%), and a -11pp decline toward base rates indicate retail speculation rather than informed pricing. My 16pp deviation from the market is justified by the ~100x gap between market price and geological reality.

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
2026-04-04	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 probability of a supervolcano eruption (VEI ≥ 8) in any given year greater than 0.05% (i.e., recurrence interval shorter than ~2000 years)? (w=0.40)
Is there currently elevated unrest (increased seismicity, ground deformation, or other precursor activity) at any known supervolcano system (Yellowstone, Toba, Campi Flegrei, Long Valley, etc.) that materially increases near-term eruption probability? (w=0.35)
Over a ~23.6-year window (2026–2050), does the cumulative probability of at least one supervolcano eruption exceed 1% given historical recurrence rates? (w=0.25)

Planner reasoning

This question asks for the probability of a VEI ≥ 8 (supervolcanic) eruption anywhere on Earth before January 1, 2050 — a ~24-year window from today. The key analytical framework is Poisson probability modeling using geological recurrence rates. **Reference class / base rates**: The geological record shows approximately 47 known VEI ≥ 8 eruptions in the last ~36 million years, giving a global rate of roughly 1 per 700,000 years. However, more accessible estimates from the Quaternary record suggest ~1 per 17,000–50,000 years globally across all supervolcano systems. The Smithsonian GVP and USGS cite 1 in 730,000 years for Yellowstone alone. **Key causal mechanisms**: Unlike many forecasting questions, supervolcano eruptions are dominated by very long-timescale geological processes. The 24-year window (2026–2050) represents an astronomically tiny fraction of typical recurrence intervals, making the base rate probability extremely low (likely 0.03%–0.14% depending on assumed recurrence). Current monitoring at Yellowstone, Campi Flegrei, and other systems would provide some advance warning but not dramatically change the base rate. **What makes this different**: The question resolution window is long enough (24 years) that near-term unrest at Campi Flegrei (which has been elevated since 2012) deserves attention as a potential outlier. Campi Flegrei's most recent unrest episodes have been VEI 4–5 scale historically, not supervolcanic (VEI 8), so even a Campi Flegrei eruption likely would not qualify unless it reaches its rare supervolcanic scale. **Temporal analysis**: For YES to resolve, a VEI ≥ 8 eruption would need to occur in the next ~8,646 days. Given annual probabilities of ~0.001%–0.006%, and no current precursor activity indicating imminent supervolcanic eruption at any monitored system, the sequence of events required (sudden magmatic system destabilization at one of ~20 known supervolcano systems, escalating to a full VEI 8 eruption with little forewarning) is extremely unlikely in this timeframe. The market price should be very low — likely 0.1%–0.5%.

2. Research Parallel tool calls · legacy v2

Tool	Status	Time	Items
wikipedia_lookup	OK	1.6s	-
article_search	OK	0.5s	-
web_search	OK	64.7s	-
kalshi_data	OK	1.3s	-
code_execution	OK	0.0s	-

3. Evidence Synthesis Sonnet · 11 items · legacy v2

#	Strength	Credibility	Direction	Source	Claim	Priced In
1	STRONG	85	↓ DOWN	code_execution	Using the most optimistic global recurrence interval of ~17,000 years, the annual probability of a VEI≥8 eruption is ~0.006% (0.000059), well below the 0.05% threshold; under longer recurrence estimates (50k–730k years), it is even lower.	Yes
2	STRONG	80	↓ DOWN	wikipedia_lookup	Yellowstone's most recent supereruption was ~630,000 years ago, with prior cycles at ~1.3 Ma and ~2.1 Ma, implying a per-system recurrence on the order of 700,000+ years — far longer than the ~2,000-year threshold in sq1.	Yes
3	MODERATE	72	↓ DOWN	article_search	All recent volcanic eruption news concerns non-supervolcanic events (Kilauea, Etna, Iceland fissures, Kamchatka), with no reports of any VEI≥8 activity or imminent supervolcanic threat anywhere on Earth.	Yes
4	STRONG	93	↓ DOWN	web_search	USGS Yellowstone Volcano Observatory April 2026 update reports background-level activity: 61 earthquakes in March (largest M1.9), alert level NORMAL, aviation code GREEN, with a pause in the north caldera rim uplift observed since July 2025.	Yes
5	MODERATE	75	NEUTRAL	web_search	Kikai caldera (Japan) supervolcano magma system is slowly recharging with newly injected magma per March 2026 research, but scientists explicitly state they 'know very little about the processes that lead to a reeruption' and cannot make near-term predictions.	No
6	WEAK	55	NEUTRAL	article_search	Echinus Geyser (a hydrothermal feature, not a magmatic indicator) resumed erupting at Yellowstone's Norris Geyser Basin in early 2026, but this is a hydrothermal phenomenon unrelated to supervolcanic eruption risk.	Yes
7	MODERATE	88	↓ DOWN	web_search	As of March 31, 2026, 40 volcanoes globally were in continuing eruption status per the Global Volcanism Program, none of which are supervolcanic systems showing VEI≥8-scale precursors.	Yes
8	MODERATE	60	↓ DOWN	web_search	No reports of elevated unrest (escalating seismicity, large-scale ground deformation) at Campi Flegrei, Toba, Long Valley, or other known supervolcano systems were found in the search results for 2025–2026.	Yes
9	STRONG	87	↓ DOWN	code_execution	Under the most favorable (shortest) global recurrence estimate of 17,000 years, the cumulative 24-year probability (2026–2050) is only ~0.14%, far below the 1% threshold; under more conservative estimates, it ranges from 0.003%–0.048%.	Yes
10	MODERATE	50	NEUTRAL	kalshi_data	The Kalshi prediction market prices this event at 20%, with a 30-day decline of 11 percentage points, implying the market is significantly overpricing relative to geological base rates (~0.14% maximum).	No
11	MODERATE	78	↓ DOWN	wikipedia_lookup	The geological record of VEI≥8 eruptions (list of large eruptions) supports an extremely low frequency of supervolcanic events in any given multi-decade window, consistent with recurrence intervals of tens of thousands to millions of years.	Yes

Information Gaps

No specific current monitoring data for Campi Flegrei (Italy) unrest levels in 2025–2026, despite its well-documented bradyseism activity since 2012
No data on Long Valley Caldera (California) current seismic or deformation status
No data on Toba (Indonesia) or Aira (Japan) supervolcano monitoring status
The Kikai caldera recharging finding is new but lacks expert commentary on what timescale recharging implies for eruption probability
Absence of any search results specifically on Campi Flegrei 2025–2026 activity is a notable gap given its historical elevated unrest
No peer-reviewed estimate of the global aggregate VEI≥8 recurrence interval accounting for all known supervolcano systems simultaneously

Key Uncertainties

The correct global VEI≥8 recurrence interval is disputed — estimates range from ~17,000 years (optimistic/aggregate) to ~730,000 years (Yellowstone alone), a 40x range that drives most of the probability uncertainty
Whether Campi Flegrei's ongoing bradyseism (not captured in current search results) has materially changed since 2024 and whether it could escalate to supervolcanic scale (historically it has not)
Kikai caldera magma recharge: the significance of 'newly injected magma' for near-term eruption probability is explicitly acknowledged as unknown by scientists
The Kalshi market at 20% is dramatically higher than geological base rates suggest — this disconnect may reflect speculative retail trading, thin liquidity, or a fundamentally different interpretation of the question's resolution criteria
Whether a previously unknown or poorly monitored supervolcano system could show rapid escalation without adequate warning

Market Context

Price: 0.2 · 7d: -0.028 · 30d: -0.11
Volume: LOW-MODERATE · Depth: SHALLOW
Reliability: LOW — The market price of 20% is dramatically inconsistent with geological base rates (~0.1%–0.14%). The wide historical price range (5%–88%) and modest volume (154 contracts/day) suggest thin liquidity, speculative pricing, and possible retail noise rather than informed forecasting. The sharp 30-day decline (-11pp) may reflect some mean-reversion toward base rates, but the current price remains ~100–200x above geological estimates.

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

Independent estimate: 2% (76% confidence)

base rate

75% conf

evidence driven

78% conf

contrarian

75% conf

Spread

1pp

Low disagreement

sq1: Is the annual probability of a supervolcano eruption (VEI ≥ 8) in any given year greater than 0.05%?

Even the most optimistic global recurrence estimate (~17,000 years) yields an annual probability of ~0.006%, an order of magnitude below the 0.05% threshold. More conservative estimates (50k-730k years) put it at 0.0001%-0.002%. The threshold would require a recurrence interval shorter than 2,000 years, which is not supported by any geological evidence. Small residual uncertainty for exotic interpretations or unknown systems.

sq2: Is there currently elevated unrest at any known supervolcano system that materially increases near-term eruption probability?

10%

USGS reports Yellowstone at NORMAL/GREEN with background activity. No evidence of escalating unrest at major supervolcano systems in search results. However, Kikai caldera shows magma recharge (NEW evidence), and Campi Flegrei has ongoing bradyseism since 2012 (information gap in current data). 'Materially increases near-term eruption probability' is a low bar — Campi Flegrei's documented unrest could qualify even if it doesn't indicate imminent VEI-8 eruption. Setting at 10% to reflect that some elevated activity exists but evidence for material risk increase is weak.

sq3: Over a ~23.6-year window (2026–2050), does the cumulative probability of at least one supervolcano eruption exceed 1%?

Code execution shows cumulative 24-year probability of ~0.14% under the most optimistic recurrence (17,000 years), and 0.003%-0.048% under more conservative estimates. All well below 1%. Small residual probability for the case where the true global aggregate recurrence is shorter than commonly cited (e.g., if some scholars argue ~2,400 years aggregate, which would give ~1% over 24 years).

Show full reasoning trace

base rate: VEI≥8 eruptions occur roughly once per 17,000-730,000 years globally. Over a 24-year window, this gives a cumulative probability of 0.003%-0.14%. No VEI≥8 eruption has occurred in recorded human history (last ~26,500 years, Taupo Oruanui).

evidence updates: All current monitoring evidence pushes DOWN or NEUTRAL: Yellowstone at NORMAL alert, no Campi Flegrei escalation reports, no known supervolcano system showing VEI-8 precursors. Kikai recharge is mildly notable but explicitly not predictive on near-term timescales. Resolution ambiguity (whether 'supervolcano erupts' requires strict VEI-8 or just any eruption from a supervolcano system) is the main reason to bump above pure geological base rate.

combination method: The sub-questions are diagnostic threshold questions rather than direct probabilities of the event. I anchor primarily on sq3 (the direct geological calculation) and adjust slightly upward for resolution ambiguity and unknown systems.

final: Strict VEI-8 interpretation gives ~0.1-0.5%. Adding resolution ambiguity premium (the market might count a large but sub-VEI-8 eruption from a supervolcano system) and uncertainty about lesser-monitored systems brings it to ~2%. Final: 0.02.

ensemble: {'ensemble_count': 3, 'perspectives': {'base_rate': {'p_yes': 0.02, 'confidence': 0.75}, 'evidence_driven': {'p_yes': 0.02, 'confidence': 0.78}, 'contrarian': {'p_yes': 0.03, 'confidence': 0.75}}, 'spread': 0.01, 'individual_p_yes': {'base_rate': 0.02, 'evidence_driven': 0.02, 'contrarian': 0.03}}

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

Overconfidence Missing Info

Challenges

Resolution criteria ambiguity is underweighted. The question asks 'When will a supervolcano next erupt?' — the term 'supervolcano' is contested. If the market resolves YES on any eruption from a supervolcano *system* (e.g., a small phreatic event at Campi Flegrei or Yellowstone hydrothermal activity), the probability is dramatically higher than a strict VEI≥8 interpretation. The forecaster acknowledges this but only adds a tiny premium (~1.5pp). Given an 8,646-day (23.6-year) window, even modest non-VEI-8 activity at any of ~20 supervolcano systems could resolve YES under a loose interpretation.
The 23.6-year time horizon is long enough that information gaps matter more than the forecaster credits. No current data on Campi Flegrei (which has had documented unrest since 2012), Long Valley, Toba, or Aira represents real uncertainty that should widen the distribution upward, not be dismissed.
The forecaster's confidence (0.76) seems high given (a) ambiguous resolution criteria, (b) acknowledged information gaps on multiple major systems, and (c) a 24-year horizon. A more honest treatment would lower confidence and slightly widen the probability.
While the market at 20% is likely overpriced due to thin liquidity, the 30-day decline of -11pp suggests some informed participants are pushing toward base rates. The forecaster correctly identifies the market as noisy, but going all the way to 2.3% may be overconfident given resolution ambiguity — a range like 3-6% better captures the ambiguity premium.

Suggested adjustment: +2pp

Modest upward adjustment (to ~4%) is warranted not because of market price, but because the resolution-ambiguity premium for what counts as a 'supervolcano eruption' over a 23.6-year window is underweighted, and information gaps on Campi Flegrei, Long Valley, Toba, and Aira are non-trivial. The geological VEI-8 base rate of ~0.1% is correct, but the question's loose phrasing and long horizon justify a larger ambiguity buffer than the forecaster applied.

Market comparison: Forecaster: 2.3%, Market: 20%. The market is almost certainly overpriced relative to geological reality and likely reflects retail speculation, thin liquidity (154 contracts/day, shallow orderbook), and definitional confusion. The 30-day decline (-11pp) suggests mean reversion is underway. The forecaster's direction is correct; the divergence is largely justified. However, the forecaster's point estimate may be slightly too low due to underweighting resolution ambiguity over a 24-year window.

6. Aggregator Opus · final calibrated forecast

4% (70% confidence)

7. Entry Decision Opus Trader

Side

Position Size

$1500

Contracts

1875

Trader Confidence

70%

Trade Rationale

Strong 21pp edge with high forecaster confidence (0.70). Geological base rates for VEI≥8 eruptions over a 24-year window are well-established at <1%, and even with a generous interpretation allowing for 4% p_yes, the market's 25% is dramatically mispriced. The forecaster's reasoning is sound and well-anchored to base rates, with the Devil's Advocate concerns (resolution ambiguity) already incorporated via the upward adjustment from 2% to 4%. The main risks are the very long horizon (23+ years to resolution, capital tied up) and resolution ambiguity, which prevent maximum sizing.

Allocation Logic

Sized at $1500 (above $1000 baseline) due to the large 21pp edge and high confidence, but capped below max because of the extreme time horizon (8646 days) tying up capital and the legitimate resolution ambiguity flagged by the critic.

Entry price: $0.80

Current: $0.73

Status: OPEN

P&L: -$131.25