← Back to scans

When will a supervolcano next erupt?

KXERUPTSUPER-0-50JAN01 · Climate and Weather · 2026-04-04
4%
Agent
32%
Market Price
-27.5%
Edge
50%
Confidence
Final Rationale
My independent estimate of ~1% was adjusted upward after the Devil's Advocate correctly identified that the resolution criteria are ambiguous — 'supervolcano eruption' could mean any eruption from a supervolcano system (not strictly VEI-8). Under strict VEI-8 interpretation, the probability is ~0.01% over 24 years based on USGS rates of 1/730,000 per year. Under a broader interpretation, Campi Flegrei's unprecedented unrest (strongest in centuries) gives perhaps 5-15% chance of some eruption, pushing the blended estimate to ~4%. The market at 30% with extremely shallow orderbook (5%-88% historical range) and low volume is a very weak signal — I deviate by 26 percentage points because even the most generous interpretation of resolution criteria doesn't support 30%, and the market appears dominated by retail mispricing of rare geological events.
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 $t{=}0$.
$n = 50$ follow-ups
1. Planner Opus · Fermi decomposition + tool plan
Tools chosen: wikipedia_lookup article_search web_search kalshi_data code_execution
Sub-questions (Fermi decomposition)
  1. Is the annual base rate probability of a supervolcano eruption (VEI ≥ 8) in any given year above 0.01%? (w=0.40)
  2. Are any of the world's major supervolcano systems (Yellowstone, Toba, Campi Flegrei, Long Valley, etc.) currently showing elevated unrest that meaningfully raises near-term eruption probability above baseline? (w=0.40)
  3. Given ~24 years remaining until resolution, does the cumulative probability of a supervolcano eruption before Jan 1, 2050 exceed 1%? (w=0.20)
Planner reasoning
This question requires estimating the probability of a supervolcano (VEI ≥ 8) eruption over the next ~24 years. The key analytical challenge is that supervolcano eruptions are extremely rare geological events — the reference class is the geological record spanning millions of years, not human historical memory. The critical steps are: (1) Establish the geological base rate for VEI ≥ 8 eruptions globally (roughly 1 per 50,000–100,000 years by most estimates, though this is debated); (2) Check whether any specific caldera systems are showing elevated unrest (Campi Flegrei in Italy has received particular attention recently for seismic swarms; Yellowstone is routinely monitored); (3) Apply the base rate over 24 years to get cumulative probability. The key difference from other rare-event questions is that even elevated unrest in a caldera rarely indicates imminent supereruption — volcanologists often note that Yellowstone's last 'supereruption' was ~640,000 years ago and it is not statistically 'overdue' in a meaningful sense. The probability is almost certainly below 1%, likely in the 0.02%–0.1% range for the full 24-year window, making this a very low probability YES outcome.
2. Research Parallel tool calls · legacy v2
ToolStatusTimeItemsSummary
wikipedia_lookup OK 1.6s -
article_search OK 0.6s -
web_search OK 59.7s -
kalshi_data OK 1.4s -
code_execution FAILED 0.0s -
Errors: code_execution: __import__ not found
3. Evidence Synthesis Sonnet · 12 items · legacy v2
#StrengthCredibilityDirectionSourceClaimPriced In
1 STRONG - ↓ DOWN web_search USGS officially estimates the annual probability of a super-eruption (VEI ≥ 8) at 1 in 730,000 (~0.00014%), well below the 0.01% threshold in the sub-question. Yes
2 STRONG - ↓ DOWN wikipedia_lookup The most recent confirmed VEI-8 eruption globally was the Oruanui eruption of Taupō Volcano approximately 25,600 years ago, consistent with extremely long recurrence intervals. Yes
3 MODERATE - NEUTRAL wikipedia_lookup Geological record of VEI-8 eruptions is incomplete due to erosion and plate tectonics, meaning the true base rate is uncertain and could be somewhat higher than recorded frequency suggests. Yes
4 STRONG - ↑ UP article_search Campi Flegrei supervolcano near Naples is showing its strongest seismic activity in centuries, with two recent earthquakes described as 'the strongest we've ever seen,' forcing evacuations in 2025. Yes
5 MODERATE - NEUTRAL web_search Campi Flegrei, Yellowstone, Long Valley, and Toba are all described as being in a 'resurgence stage' with ground deformation, earthquake swarms, and small-scale eruptions being common features — but these do not necessarily indicate imminent supereruption. Yes
6 STRONG - ↓ DOWN web_search USGS states Yellowstone is not overdue for an eruption and scientists are not convinced another catastrophic eruption will ever happen; most scientists believe precursors would be detectable weeks to months in advance. Yes
7 MODERATE - NEUTRAL article_search Yellowstone's Echinus Geyser resumed erupting in March 2026 after five years of silence — a routine hydrothermal event, not a volcanic unrest indicator of supereruption significance. No
8 STRONG - NEUTRAL article_search Multiple non-supervolcano eruptions occurred in 2025 (Kilauea, Etna, Iceland Reykjanes, Kanlaon) — these are VEI 1–4 events unrelated to supervolcano systems and do not elevate supereruption risk. Yes
9 STRONG - ↓ DOWN web_search At the USGS annual rate of 1 in 730,000 (~0.00014%), the cumulative probability over 24 years is approximately 0.0034% — roughly 300 times below the 1% threshold. Yes
10 WEAK - ↑ UP kalshi_data The Kalshi prediction market for this question (KXERUPTSUPER-0-50JAN01) is currently priced at 30%, with a +5% 7-day change and +8% 30-day change, implying far higher probability than scientific base rates. No
11 WEAK - NEUTRAL kalshi_data The Kalshi market has traded as low as 5% and as high as 88%, suggesting extreme price volatility and thin/unsophisticated trading — the 30% price is likely not a reliable probability estimate. No
12 MODERATE - NEUTRAL web_search A new study shows ancient supervolcano systems remain hazardous for thousands of years post-eruption, suggesting prediction models may need revision — but this does not materially change near-term probability estimates. No
Information Gaps
  • No quantitative volcanological assessment of how much Campi Flegrei's current unrest actually raises near-term supereruption probability above baseline (most coverage is qualitative)
  • No data on the specific VEI magnitude the Campi Flegrei system is capable of in a next eruption — it may not reach VEI-8 even if it does erupt
  • No updated USGS or academic paper refining the global VEI-8 recurrence rate beyond the ~1/730,000 per year figure
  • No information on current monitoring status or unrest levels at Toba, Long Valley, or other major supervolcano systems beyond Yellowstone and Campi Flegrei
  • No explanation for why the Kalshi market is priced at 30% — unclear if it reflects question misinterpretation, thin liquidity, or actual informed beliefs
Key Uncertainties
  • Whether Campi Flegrei's elevated seismic activity represents a genuine precursor phase or a non-eruptive bradyseism episode (historically, most Campi Flegrei unrest episodes have not resulted in eruptions)
  • The true geological base rate of VEI-8 eruptions globally — if the record is systematically undercounted, actual rates could be 2-5x higher than estimated, though still far below 0.01%/year
  • Whether the Kalshi market price of 30% reflects genuine uncertainty (perhaps about question resolution criteria) or is simply mispriced due to low volume and retail participation
  • The question's resolution criteria — whether 'supervolcano eruption' requires VEI-8 or could include any eruption from a supervolcano system (which would dramatically change probability)
Market Context
Price: 0.3 · 7d: +0.05 · 30d: +0.08
Volume: LOW — 149 contracts/day average is very thin for a binary market · Depth: SHALLOW — wide price range (5%–88%) suggests poor liquidity and high volatility
Reliability: LOW — extreme price range and low volume indicate this market is not reliably pricing scientific probability; the 30% price is almost certainly a significant overestimate relative to geological base rates, possibly due to retail misunderstanding of VEI-8 threshold or question scope ambiguity
4. Independent Forecaster Opus · No market price · legacy v2 Did NOT see market price
Independent estimate: 1% (high confidence)
sq1: Is the annual base rate probability of a supervolcano eruption (VEI ≥ 8) in any given year above 0.01%?
3%
The USGS estimate of ~1 in 730,000 per year (0.00014%) is well below the 0.01% threshold (1 in 10,000). Even accounting for incomplete geological records that might push the true rate 2-5x higher, we'd still be at ~0.0003-0.0007%, which is still an order of magnitude below 0.01%. The geological record of VEI-8 events (roughly 47 identified in the last 36 million years, or about 1 per 766,000 years) is consistent with the USGS figure. The only way this sub-question resolves YES is if the base rate is systematically undercounted by a factor of 50-100x, which seems extremely unlikely given modern geological survey capabilities. I assign ~3% probability.
sq2: Are any of the world's major supervolcano systems currently showing elevated unrest that meaningfully raises near-term eruption probability above baseline?
8%
Campi Flegrei is genuinely showing its most significant unrest in decades/centuries, with strong seismic swarms and ground uplift. This pushes UP from baseline. However, several critical factors push DOWN: (1) Campi Flegrei has experienced multiple bradyseism episodes (notably 1970s-1980s) without eruption; (2) Even if Campi Flegrei erupts, its next eruption is most likely to be VEI 3-5, not VEI 8 — its last VEI-8 was the Campanian Ignimbrite ~39,000 years ago; (3) Yellowstone is explicitly not showing concerning signs per USGS; (4) Volcanological consensus is that caldera unrest episodes overwhelmingly do not lead to supereruptions. The question asks if unrest 'meaningfully raises' probability — Campi Flegrei's unrest probably raises eruption probability for that system by some factor, but raising it to a level that materially affects the global VEI-8 probability is a much higher bar. I estimate ~8% that current unrest meaningfully raises near-term VEI-8 probability above baseline.
sq3: Given ~24 years remaining until resolution, does the cumulative probability of a supervolcano eruption before Jan 1, 2050 exceed 1%?
1%
At the USGS rate of 1/730,000 per year, 24 years gives cumulative probability of ~0.003%. Even if I use a more generous rate of 1/100,000 per year (which would be at the very high end of any credible estimate), 24 years gives ~0.024%, still 40x below 1%. To reach 1% cumulative probability over 24 years, you'd need an annual rate of about 1/2,400, which is roughly 300x higher than the USGS estimate and completely inconsistent with the geological record. The only scenario where this exceeds 1% is if we're in a fundamentally different geological regime than the past millions of years suggest, or if the question resolves on something other than VEI-8. I assign ~1% probability.
Show full reasoning trace
base rate: VEI-8 eruptions occur roughly once every 500,000-1,000,000 years globally. USGS estimates 1/730,000 per year. Over 24 years, this gives a cumulative probability of approximately 0.003%.
evidence updates: Campi Flegrei unrest pushes up slightly but: (1) most unrest episodes don't produce eruptions, (2) even if it erupts it likely won't be VEI-8, (3) one system's unrest doesn't dramatically change global VEI-8 probability. Yellowstone is explicitly not concerning. Net effect: maybe 2-5x multiplier on base rate, bringing us to ~0.007-0.017%. The note about a Kalshi market at 30% is almost certainly mispriced or reflects misunderstanding of 'supervolcano eruption' (perhaps traders think any eruption from a volcano that has historically been 'super' counts, or it's a thin/illiquid market).
combination method: Direct probability estimation based on geological base rates with uncertainty adjustments, rather than mechanical combination of sub-question weights (which were structured to evaluate components of the estimate).
final: The probability of a VEI-8 supervolcano eruption before January 1, 2050 is extremely low — on the order of 0.01-0.05%. I set my estimate at 0.5% (0.005) to account for: deep uncertainty in geological records, the possibility that the question might resolve on criteria slightly broader than strict VEI-8, unknown unknowns, and the general principle of not being overconfident on very low probability geological events. This is still very low but acknowledges meaningful model uncertainty.
5. Devil's Advocate Sonnet · legacy v2 Market price revealed
Reasoning Flaw Missing Info Math Error Asymmetric Evidence
Challenges
  1. The forecaster's final estimate of 1% (0.010) is inconsistent with their own reasoning trace, which concludes at 0.5% (0.005). The combined p_yes should match the reasoning trace conclusion. This appears to be a transcription or rounding error.
  2. The forecaster should more carefully consider the resolution criteria. The question asks 'When will a supervolcano next erupt?' — this may not strictly require VEI-8. If the question resolves on any eruption from a caldera system commonly labeled a 'supervolcano' (e.g., Campi Flegrei producing a VEI-4 or VEI-5), the probability would be meaningfully higher. The forecaster acknowledges this possibility ('the question might resolve on criteria slightly broader than strict VEI-8') but doesn't adequately investigate the actual resolution criteria.
  3. The 2-5x multiplier for uncertainty seems arbitrary and insufficiently justified. If the geological record is incomplete by a factor of 2-5x (as the forecaster acknowledges is possible), AND there's question scope ambiguity that could include VEI-6 or VEI-7 events, these multiplicative uncertainties could compound. A VEI-7 event has a much higher base rate (~1/1,000-5,000 years) and would drastically change the calculation if it counts.
  4. The forecaster dismisses the Kalshi market price as 'almost certainly mispriced' but then notes 'unclear if it reflects question misinterpretation' — this is a critical unresolved question. If market participants are correctly interpreting the resolution criteria to include sub-VEI-8 eruptions from supervolcano systems, then the market could be much more reasonable than the forecaster assumes. The forecaster should have investigated the actual resolution criteria more carefully before dismissing the market signal.
  5. The forecaster's treatment of Campi Flegrei is somewhat dismissive. While it's correct that most unrest episodes don't produce eruptions, the current episode is described as the 'strongest seismic activity in centuries.' If Campi Flegrei were to erupt at any scale, this could potentially satisfy the question's resolution criteria depending on how 'supervolcano eruption' is defined — and the probability of *some* eruption from Campi Flegrei in the next 24 years given current unrest is non-trivial (perhaps 5-15%).
  6. With 8672 days (~23.7 years) until resolution, there is substantial time for conditions to change at multiple volcanic systems. The forecaster's analysis is essentially static — it doesn't account for the possibility that new unrest could develop at any of dozens of large caldera systems over a 24-year window.
Suggested adjustment: +3pp
The primary reasoning flaw is insufficient investigation of the resolution criteria. If 'supervolcano eruption' means any eruption from a supervolcano system (not strictly VEI-8), the probability is substantially higher. Campi Flegrei alone has perhaps a 5-15% chance of *some* eruption in 24 years given current unrest, and there are other systems too. Even if the question requires VEI-7+, the base rate is orders of magnitude higher than VEI-8. The internal inconsistency between 0.5% in reasoning and 1.0% in the final answer also needs resolution. Adjusting upward by 3 percentage points to ~4% to account for resolution criteria ambiguity, while still staying far below the likely mispriced market.
Market comparison: Forecaster: 1.0%, Market: 30.0%. The 29-point divergence is enormous. The market is LOW volume (149 contracts/day) and has extreme historical range (5%-88%), making it unreliable as a price signal. However, the divergence is large enough that the forecaster should seriously consider whether market participants understand the resolution criteria differently — specifically, whether any eruption from a 'supervolcano' system counts, not just VEI-8. If so, 30% could be defensible (especially given Campi Flegrei's unrest). If the question strictly requires VEI-8, the market is almost certainly wildly overpriced and the forecaster's low estimate is justified.
6. Aggregator Opus · final calibrated forecast
4% (medium confidence)
My independent estimate of ~1% was adjusted upward after the Devil's Advocate correctly identified that the resolution criteria are ambiguous — 'supervolcano eruption' could mean any eruption from a supervolcano system (not strictly VEI-8). Under strict VEI-8 interpretation, the probability is ~0.01% over 24 years based on USGS rates of 1/730,000 per year. Under a broader interpretation, Campi Flegrei's unprecedented unrest (strongest in centuries) gives perhaps 5-15% chance of some eruption, pushing the blended estimate to ~4%. The market at 30% with extremely shallow orderbook (5%-88% historical range) and low volume is a very weak signal — I deviate by 26 percentage points because even the most generous interpretation of resolution criteria doesn't support 30%, and the market appears dominated by retail mispricing of rare geological events.
7. Entry Decision Opus Trader
Side
NO
Position Size
$1000
Contracts
1460
Trader Confidence
-
Entry price: $0.69
Current: $0.79
Status: ARCHIVED
P&L: —