superthesis
superthesis × polymarketmarket watch · 01
volcano season still has six months

The crowd says 2026 stays quiet.
The record says otherwise.

A VEI 4-free calendar year would be unprecedented in 25 years of essentially complete global records. H1 is clean, but six months of active volcanic season remain across dozens of monitored systems.

A price tells you what people bet. A verdict tells you why.
analysis as of Jun 24, 2026 · crowd price as of Jun 24, 2026
/ the questionmarket snapshot · Jun 24, 2026
/ what the market is pricing

One claim. Real money on the line.

The claim under test
Will there be exactly 0 confirmed VEI 4 or higher volcanic eruptions worldwide in 2026?
$498K tradedresolves 2026-12-31sciencepolymarket
Polymarket crowd · Jun 24, 2026
68%
implied probability · yes
0 · nolong shotyes · 100
superthesis interval0.12 (0.09–0.15)
/ superthesis · run on the same claimthe engine, not the odds
/ argue both sides, score the evidence

What the engine actually finds.

superthesis · adversarial verdict verdict · 0.09–0.15
“Will there be exactly 0 confirmed VEI 4 or higher volcanic eruptions worldwide in 2026?”
The crowd gives this ~69%; after arguing both sides, superthesis lands far lower — about 12%.
Mayon 2026 confirmed VEI 1-2 by column height (~1.3 km above summit, well below the ~10 km VEI 4 threshold)
cleared
VOTW v5.3.6 (16 May 2026) catalogues 47 confirmed 2026 eruptions with no VEI 4 designation
cleared
25-year GVP record shows at least one VEI 4+ event in every year from 2000 through 2024
refuted
H2 2026 global volcanic activity across all monitored systems
open
GVP finalized 2026 dataset and any retroactive VEI reclassifications (expected by March 2027)
open
Thesis · the case for
H1 2026 is confirmed clean: Mayon's eruption is firmly VEI 1-2, and VOTW v5.3.6 lists 47 active eruptions with no VEI 4 designation. Conditional on a quiet H1, a Poisson model at the observed base rate yields roughly 47% probability that H2 also sees zero qualifying events — placing the full-year outcome within reach of the Poisson tail.
Antithesis · the case against
The GVP's own FAQ states VEI 4 frequency has been impressively constant for more than a century, and the 21st-century eruption record shows at least one confirmed VEI 4+ event in every year from 2000 through 2024 — a 25-year unbroken streak with no zero-count year to calibrate against. An unconditional Poisson model places the probability of a zero-count year at only ~22%, and the remaining six months of H2 are statistically consistent with base-rate expectations, not anomalous quiescence.
Synthesis · the calibrated reading
The claim that 2026 will record exactly zero confirmed VEI 4+ volcanic eruptions is almost certainly false. The decisive evidence is the uncontested 25-year empirical record (2000–2024) showing at least one confirmed VEI 4+ eruption start in every year, established by T1 GVP sources, with a base rate the GVP itself calls 'impressively constant for more than a century'; a Poisson model yields only ~22% probability of a zero year unconditionally, and no zero-count year exists in the modern complete record to calibrate against. While H1 2026 does appear clean of VEI 4 events — Mayon's effusive eruption is firmly characterized as VEI 1–2 by column height data — this is statistically consistent…
0.12superthesis signal
0.09–0.15
0 · refutedcontestedconfirmed · 1.0
// the cases, gates, and the calibrated signal are from a live superthesis run on this claim. every source it weighed is listed below.
/ crowd vs. enginea price is not an argument
/ why this matters

The market gives you a number.
We give you the reasoning.

ContrarianWe read 57 pts lower than the crowd, so we flip the crowd’s call — the market leans one way, we lean the other. On conviction, our stance is more resolute — 19 pts further from a coin-flip.
/ same claim · two ways of knowing
Polymarket · 0.68
superthesis · 0.12
0 · won't happen1.0 · will happen

A prediction market aggregates bets; superthesis aggregates arguments — it decomposes the claim, makes the strongest case on each side from sourced evidence, and reports a calibrated reading you can audit. The price says where the crowd sits; the verdict shows why.

/ what $1,000 does · estimated at the current price
Our call
No the outcome our reading favors
$1,000 returns if our call is right
$3,175 profit $2,175
Market price vs our read
31¢ market · we read 88¢ — we read our call cheaper than the market
Expected value if our read is right
+$1,794position worth ~$2,794 — above the $1,000 stake

Illustrative only — not financial advice and not a recommendation to bet. The return is an estimate at the headline price for a $1,000 position on our call; the “expected value” assumes our calibrated reading is the true probability — exactly what’s uncertain. Prices move; the live market is linked above.

/ since we called itthe record, live
/ has the crowd moved our way since Jun 23, 2026?

Since we called it — where the crowd went.

Our reading · Jun 23, 2026
0.12
Crowd · Jun 23, 2026
68%
Crowd · Jul 17, 2026
68%
Held
our read 0.12 then now
0 · won’t happen1.0 · will happen

The crowd has barely moved (±1 pt) since — too early to tell.

// live Polymarket price, keyed to this market’s id and re-fetched when you load the page. “Toward” on an open market means the crowd has drifted our way — suggestive, not proof; being early isn’t being right until it resolves.

Stop reading the odds. Read the evidence.

Point superthesis at any claim, market, or question — and watch it argue both sides, cite every source, and resolve a calibrated signal. Free while The Field is in early access.