Skip to content

Summary

Research scan (2024–2026) across academic papers, central-bank/SEC working papers, and market-structure news, filtered for what changes the framework. The headline: the unconditional premium-selling edge that TOMIC assumed is no longer supported by the evidence — the edge is now conditional, which vindicates the framework's regime-gating and baseline-comparison design while raising the bar on what counts as edge.

1. The Variance Risk Premium Is Conditional (Highest Impact)

  • Chicago Fed WP 2025-17: option alphas have become statistically indistinguishable from zero over the past ~15 years; examines traded vs synthetic options.1
  • Dew-Becker & Giglio, "The Decline of Option Returns": identifies a structural break around 2012 — after which standard S&P 500 option strategies no longer show statistically significant alphas or information ratios.23
  • Counterweight: a 2025 study finds variance risk pricing still statistically significant through ~9-month maturities (declining with maturity).4
  • Framework consequence: "sell premium because VRP is positive on average" (TOMIC-2012) is an unsupported prior. Edge claims must be conditional (regime, tenor, structure) and must beat the RV/GARCH baselines net of costs — exactly what the EV contract demands. Day-1 rehearsal already demonstrated this gate. Candidate amendment: the regime mapping's VRP variable may need a size threshold, not just sign.

2. Algorithms and ML

  • Deep hedging: RL hedgers conditioned on the IV surface and its dynamics beat practitioner/implied-delta benchmarks in SPX experiments; work now addresses transaction costs, position limits, imitation learning.5
  • Vol forecasting: the HAR-RV bar holds — a careful large study finds ML models did not beat HAR after proper rolling-window design; hybrid RNN-HAR models show promise but no universal winner.67 → The framework's GARCH/HAR baseline requirement is the right bar, not a conservative handicap.
  • Vol surface ML: VolGAN (arbitrage-free surface generation), neural operators, meta-learning for sparse-quote reconstruction — relevant to surface features.8
  • LLM agents in trading: 2025 surveys map architectures (trading, risk, strategy generation) but emphasize the research-to-deployment gap; no evidence LLM agents outperform professionals at prediction.9 → Consistent with the framework's bounded LLM roles (triage/synthesis, never probability estimation).

3. Market Structure and Conditions

  • 0DTE is not inherently destabilizing: MM intermediation in 0DTE lowers volatility (vol attenuation, 2019–2023 intraday data); MM net gamma is on average positive and negatively related to subsequent intraday vol. The "0DTE Conundrum" paper treats liquidity distortions and hedging cascades.1011 → Softens the panic narrative; the 0DTE complex concept should be updated.
  • Retail dominance: retail >60% of options volume; ~90% of PFOF concentrated in three wholesalers; retail activity concentrated in short-dated purchases before earnings ("Levering up!", Yale; JF gamification study).1213
  • GEX as a signal is weak: backtests show gamma-exposure metrics mostly track VIX/ATM IV; error margins large; practitioner-grade evidence only.14 → Do NOT add GEX to the regime variable set without out-of-sample validation.
  • Extended hours era: Cboe received SEC approval (2026) for extended-hours single-stock options (launch July 13, 2026); near-24×5 equities trading proposed for EDGX (Dec 2026 target).[^cboe-extended]15 → Settlement/event mechanics shift again; the framework's event map must include cross-session gaps.
  • Weekly options history: difference-in-differences evidence that weekly-option introduction raised equity volatility.16
  • Current tape (Sep 2026): SPX ~7,673, VIX mid-teens, contango, wide VRP — the calm-contango regime from the day-1 rehearsal.

4. Recommended Books

Book Why Source
Colin Bennett — Trading Volatility, Correlation, Term Structure and Skew Practitioner standard for vol/correlation/skew trading; author-sanctioned free PDF trading-volatility.com PDF
Euan Sinclair — Volatility Trading (2nd ed., Wiley 2013) The quantitative vol-trading standard; EV/baseline thinking Wiley (no legit free PDF)
Marcos López de Prado — Advances in Financial Machine Learning (Wiley 2018) Backtest overfitting, purged CV — the backtest discipline source Wiley
Marcos López de Prado — Machine Learning for Asset Managers (Cambridge 2020) Theory + economic justification; anti-unsubstantiated-backtest Cambridge
Jim Gatheral — The Volatility Surface: A Practitioner's Guide (Wiley 2006) The vol-surface reference Wiley (free PDFs circulate but are copyright-flagged — use the publisher copy)

Footnote sources not in the table: all URLs in sources[] and inline footnotes below are author-hosted, central-bank, SEC, NBER, SSRN, or publisher pages.

References

  • Chicago Fed WP 2025-17: https://www.chicagofed.org/-/media/publications/working-papers/2025/wp2025-17.pdf
  • Dew-Becker & Giglio: https://www.dew-becker.org/documents/option_decline.pdf ; Dew-Becker, Giglio & Arfe (2024): https://stefanogiglio.org/papers/dew-becker-giglio-arfe-2024.pdf
  • SSRN: 0DTE intermediation 4881008 ; 0DTE Conundrum 4692190 is the VRP-maturity paper; 0DTE conundrum via SSRN search
  • arXiv: 2407.21138 (deep hedging) ; 2504.06208, 2504.05521 (HAR vs ML) ; 2406.08041 (VolGAN)
  • FEDS 2025-061 ; SEC DERA 0DTE study ; Yale SOM gamification PDF ; JF 10.1111/jofi.13285 (weekly options ; gamification)
  • Cboe: extended-hours approval (ir.cboe.com) ; 24×5 insights post ; VIX dashboards
  • BNP Paribas dispersion primer: https://globalmarkets.cib.bnpparibas/equity-dispersion-trading/

Brainstorm Hooks

  • Regime mapping amendment candidate: VRP size threshold (e.g., trade only when IV − GARCH-forecast > X) — needs quorum evidence, log as proposal
  • Update 0DTE complex with the vol-attenuation findings (draft amendment note)
  • Do not add GEX to regime variables (rejected: weak evidence)
  • Extended-hours single-stock options → new event-window rules needed for cross-session gaps

  1. Chicago Fed WP 2025-17 ↩

  2. Dew-Becker & Giglio, author-hosted PDF ↩

  3. Dew-Becker, Giglio & Arfe (2024) ↩

  4. SSRN 4692190 — VRP significance through 9-month maturities ↩

  5. arXiv 2407.21138 — deep hedging with IV surface dynamics ↩

  6. arXiv 2504.06208 / 2504.05521 — HAR vs ML comparisons ↩

  7. Federal Reserve FEDS 2025-061 ↩

  8. arXiv 2406.08041 — VolGAN ↩

  9. ACM/EMNLP 2025 LLM-agent surveys ↩

  10. SSRN 4881008 — 0DTE market-maker intermediation ↩

  11. SSRN — 0DTE Conundrum ↩

  12. Yale SOM — Levering up! ↩

  13. Journal of Finance, doi 10.1111/jofi.13285 ↩

  14. FlashAlpha GEX backtest (practitioner) ↩

  15. Cboe insights — near-24×5 EDGX proposal ↩

  16. JF — weekly options and equity volatility ↩