Policy deadline density predicts compliance-market volatility

Five independent claude-sonnet-4-6 calls, one per persona — Skeptic, Fact-Checker, Researcher, Compliance-Guard, Falsification-Auditor. Each writes its hardest objection from its own seat, paired with the methodological resolution it would accept. Run on the static catalogue spec Jun 3, 2026; a live council for any topic is at /try.

1. Skeptic #01
   Raised

   The dominant alternative explanation is that EU ETS price volatility in the 2018–2026 window is mechanically driven by natural gas price shocks (TTF front-month), not regulatory deadline density, and these two variables are spuriously correlated because peak legislative activity (CSRD trilogue finalization in late 2022, CBAM entry-into-force negotiations) coincided with the 2021–2022 European energy crisis, which independently produced 200%+ ETS price swings through fuel-switching dynamics. The top-decile deadline-density quarters in this sample are thus disproportionately drawn from the 2021–2023 energy-crisis window, meaning any predictive signal attributed to deadline density is actually capturing TTF volatility regime-switching. This would generate the observed 20%+ ETS moves and apparent ICE Endex spread widening without any causal policy-uncertainty channel.

   Resolved

   Estimate a multivariate OLS regression of 90-day forward EU ETS realized volatility (computed from ICE Endex daily settlement prices) on (i) the deadline-density score, (ii) TTF natural gas front-month realized volatility over the same lookback window (sourced from CME Group or ICE historical data), (iii) the Eurostat EU industrial production index monthly change, and (iv) EEX auction supply volume as a proxy for allowance scarcity; if the deadline-density coefficient shrinks by more than 50% or loses statistical significance (p > 0.10) after including TTF volatility, the energy-macro confound explains the pattern. As a robustness check, restrict the sample to the pre-energy-crisis period (2018–2020) where TTF volatility was subdued, and test whether the deadline-density effect survives in that subsample alone—a null result in the restricted window with significance only in 2021–2023 would confirm the spurious-correlation hypothesis.

2. Fact-Checker #02
   Raised

   EU ETS EUA December futures on ICE Endex have posted annualized realized volatility of 40–70% across 2018–2024 (calculable from ICE daily settlement series, Bloomberg EUA1 Comdty), which puts the unconditional 90-day base rate of a ≥20% absolute price move at roughly 50–65% of all rolling windows — substantially above the 35% falsification threshold and nearly overlapping the 70% support threshold before any conditioning on deadline density. Compounding this, the 2018–2026 sample spans only ~32 quarters, so the top decile contains 3–4 observations; Wilson 95% confidence intervals on any observed proportion at that sample size span ±25–30 percentage points, fully overlapping both thresholds and rendering the SUPPORTS/FALSIFIES distinction statistically void. The 8–12 bps yield-tightening claim on "CBAM-linked compliance bonds" is additionally untestable: no liquid secondary market exists for such instruments, and comparable climate-aligned corporate bonds carry bid-offer yield spreads of 20–50 bps in dealer markets, placing the claimed signal below the noise floor of any available price feed.

   Resolved

   First, compute the empirical unconditional base rate of ≥20% rolling 90-day EU ETS moves over the full available ICE Endex history (2013–2024) and set the SUPPORTS threshold at a frequency that is statistically distinguishable from that base rate at 95% confidence — likely ≥80% vs. ≤45% rather than 70%/35%; shift to rolling 30-day deadline-density windows to expand the top-decile cell to n ≥ 40 observations before applying any proportion test. For the spread-widening sub-claim, use ICE Endex published daily best-bid/best-offer settlement data (tick size €0.01, sufficient for 15–25 bps resolution at current EUA price levels without requiring proprietary L2 feeds) and replace the CBAM compliance-bond yield metric entirely with the ICE Endex EUA December-to-December calendar spread, which is liquid, tick-precise, and interpretable as a deadline-sensitive term structure signal with measurable 8-bps granularity.

3. Researcher #03
   Raised

   The dominant uncontrolled confounder is EU natural gas price volatility (TTF hub), operating through the fuel-switching mechanism: gas-to-coal substitution sets the marginal abatement cost and therefore EU ETS allowance prices, so TTF spikes mechanically drive large EUA price moves independent of any policy-deadline signal. Critically, the 2021–2023 energy crisis both generated the largest EU ETS price swings in the sample and simultaneously compressed the EU legislative calendar—CBAM phase-in dates, CSRD transposition deadlines, and associated delegated acts were all front-loaded into this window—meaning "top-decile deadline-density quarters" are nearly colinear with "top-decile TTF volatility quarters," and the deadline-density coefficient will absorb the energy-shock effect, producing severely upward-biased estimates of the policy-calendar mechanism.

   Resolved

   Add front-month TTF natural gas futures (log-differenced, sourced from ICE Endex tick data or Eurostat series nrg_pc_g_m as a monthly proxy) and the clean-dark spread as time-varying covariates in the predictive regression; quarter-of-year fixed effects will additionally absorb the endogenous EU ETS compliance calendar (April surrender deadline seasonality). To identify the deadline-density effect causally rather than conditionally, instrument the density variable using legislatively predetermined procedural milestones—EU trilogue conclusion dates and SEC final-rule publication lags extracted from EUR-Lex procedure files and SEC EDGAR rulemaking dockets—since these upstream procedural dates are set 12–18 months in advance by institutional timetables that are plausibly orthogonal to contemporaneous TTF price realizations, providing a valid exclusion restriction.

4. Compliance-Guard #04
   Raised

   The quantified predictions embedded in this hypothesis—15–25 bps EU ETS spread widening, 50–100 bps carbon-futures outperformance, and 8–12 bps CBAM-bond yield tightening—could be incorporated into investment research or discretionary trading mandates before the ≥70% hit-rate threshold is confirmed, creating liability under MiFID II Article 24 (suitability of investment recommendations) and EU Market Abuse Regulation 596/2014 Article 20 (standards for investment recommendations), because the signal would be presented with a precision that implies empirical validation it does not yet have. Simultaneously, if a company's CSRD Article 19a sustainability report or IFRS S2 transition-risk disclosure relies on this unvalidated deadline-density metric to quantify carbon-price exposure, the resulting figures could constitute a material misstatement, exposing preparers to enforcement by national competent authorities under CSRD's mandatory assurance regime and, for cross-listed issuers, to SEC Rule 10b-5 liability for misleading forward-looking risk quantification.

   Resolved

   The hypothesis must carry an explicit gating disclaimer—"experimental signal, below SUPPORTS threshold, not suitable for investment recommendations or regulatory disclosure"—until the full 2018–2026 back-test confirms the ≥70% hit rate in top-decile deadline-density quarters, an out-of-sample hold-out period (e.g., 2024–2026) replicates the result, and EU ETS price-move calculations are independently cross-validated against ICE Endex official settlement data sourced through /api/carbon-pricing with auditable version control. Only after peer review of the deadline-calendar construction methodology and formal documentation that the signal clears the SUPPORTS threshold—and explicitly does not meet the FALSIFIES criterion—may the quantified spread and yield predictions be incorporated into MiFID II-compliant research notes or cited as a transition-risk input in CSRD/IFRS S2 disclosures.

5. Falsification-Auditor #05
   Raised

   EU ETS allowance prices have exhibited realized annualized volatility of roughly 40–55% over the 2018–2026 window, implying that a 90-day window sampled at random already has an unconditional probability of producing a ≥ 20% absolute price move on the order of 45–55% (a simple lognormal approximation at 50% annualized vol gives σ₉₀ ≈ 30%, yielding P(|r| ≥ 20%) ≈ 50%). The FALSIFIES band of ≤ 35% of top-decile quarters showing a ≥ 20% move is therefore likely below the null base rate, meaning a completely uninformative deadline-density signal would still exceed the FALSIFIES threshold most of the time — the band is practically unreachable under the null. This asymmetry makes the hypothesis unfalsifiable as currently specified: the "no edge over base rate" criterion is set below the actual unconditional base rate of large moves.

   Resolved

   First, compute the empirical unconditional base rate of ≥ 20% EU ETS moves in all 90-day rolling windows from 2018–2026, along with a bootstrap 95% confidence interval around that base rate; the FALSIFIES threshold must be set below this interval's lower bound to be distinguishable from the null. Second, run a label-permutation Monte Carlo (≥ 10,000 iterations) in which deadline-density quintile assignments are randomly shuffled across quarters while the price-move series is held fixed, producing a null distribution of top-decile hit rates; the FALSIFIES band should be redefined as the 5th percentile of this null distribution (likely closer to 38–44%), and the SUPPORTS band should require the observed hit rate to exceed the 95th percentile, converting the binary threshold test into a proper two-sided permutation test with explicit, reachable rejection regions on both sides.

Unlike the static stress tests above (synthesised against the frozen catalogue spec), this is what a 3-voice council found in the most recent biweekly review. Refreshed on the 1st and 15th of each month at 09:00 UTC. Each voice runs one bounded web search via Anthropic's web_search_20260209 tool, cites what it finds, and recommends a verdict. The verdict diverges from the curated catalogue status (forming) — the synthesis below explains why.

Agent draft incorporating the 9 cited findings from the live council above. Not auto-merged — surfaces here for human review. To accept, open a PR editing site/src/_data/hypotheses.json with the revised fields below. To reject, ignore and the proposal will refresh on the next council run.

1. forming

   May 30, 2026 · added to catalogue at status "forming"

• ICE Endex EUA December-contract market-making desks widen posted bid-ask spreads 15-25 bps versus the trailing 60-day baseline during top-decile regulatory-deadline-density windows (CSRD filing gates, CBAM quarterly reporting deadlines), compressing order-book depth by an estimated 18-30% within the 30-day pre-deadline cluster.
• Systematic carbon-trading desks at Hartree Partners and Vitol Carbon running deadline-density-conditioned long-EUA overlays outperform passive EUA benchmark holdings by 50-100 bps on 90-day positioning windows aligned to CSRD and CBAM phase-in milestones, with signal alpha concentrated in the 45 days immediately preceding each deadline.
• ArcelorMittal and HeidelbergMaterials, the two largest CBAM-liable EU importers by embedded-carbon tonnage, see CBAM certificate forward prices tighten 8-12 bps in the 60-90 days following each major CBAM quarterly reporting deadline as compliance-quantity uncertainty resolves into verified declarations.
• HSBC Climate Bond desk and BNP Paribas Green Bond origination teams are required to embed a 5-10 bps deadline-density uncertainty premium into EU Taxonomy-aligned green bond spread models by the 2026 CSRD full-scope reporting phase, affecting pricing on an estimated €80-120B in annual European green bond issuance.
• ESMA's EMIR volatility-surveillance mandate triggers a formal consultation on EUA derivative position-limit tightening within 24 months of the first peer-reviewed validation of the deadline-density signal, with proposed speculative net-position caps set at 15-20% of open interest for non-hedging counterparties.