Volatility Trading and Options Market Making
Volatility trading and options market making form a large and rapidly evolving segment of global financial markets. Options market makers provide continuous two-sided liquidity across single-stock, index, ETF, and cryptocurrency options, earning the bid-ask spread while managing the inventory risk embedded in the options they hold. Their hedging activity — particularly the gamma-driven programmatic rebalancing that dampens or amplifies price moves depending on the sign of aggregate dealer gamma — has become a primary determinant of short-term equity price distributions, at times outweighing fundamental corporate data. Volatility traders, by contrast, take directional or relative-value views on the level of implied or realized volatility using options, variance swaps, VIX futures, and related instruments.
Recent Developments
The options market making industry has undergone a structural transformation through 2025–2026. Retail participation has reached unprecedented levels, with zero-days-to-expiry (0DTE) options now representing over 60 percent of S&P 500 options volume, creating new regimes of intraday gamma exposure where market maker hedging algorithms operate at compressed time horizons. In February 2026, 0DTE options reached a record 63 percent of SPX volume on a record monthly SPX average daily volume of 4.75 million contracts. In March 2026, institutional investors accounted for 48 percent of 0DTE activity as they stepped up hedging amid Middle East conflict fallout, while non-0DTE S&P 500 options volume rose 27 percent, pushing total S&P 500 options average daily volume to an all-time high of 5.3 million contracts.[^c1][^c2] O'Donovan (2026) provides causal evidence that 0DTE options compressed SPX put skew across all tenors, peaking at 71 basis points at 30 days, reducing the annual cost of equity tail insurance by approximately $1.4 billion through substitution and dealer warehousing channels. Chong and Todorov (2026, Journal of Finance) derive tight pricing kernel restrictions from 0DTE options, finding no feasible pseudo-arbitrage after accounting for transaction costs, estimation risk, and short-term volatility risk. The June 18, 2026 quadruple witching was the largest options expiration in history at $8.3 trillion notional, while the SpaceX options debut on June 17 set records with 1.8 million contracts and $2.8 billion in premium on day one. Nearly a month later, SpaceX continued to trade at an implied volatility of 92, almost 3.5 times that of QQQ, with persistently call-skewed options flow.
In market structure, the SEC proposed the rescission of Regulation NMS Rule 611 (the Order Protection Rule) on June 11, 2026, representing the most significant equity market structure reform in two decades and signaling a broader deregulatory shift under Chairman Paul Atkins. Cboe launched extended trading hours for select single-stock options on July 13, 2026, offering pre-market (7:30 a.m. ET) and post-market (4:00–4:15 p.m. ET) trading for approximately 20 highly liquid names including the Magnificent Seven. Cboe also launched Cboe Predicts, its prediction markets suite featuring XSP binary options cleared through the OCC, with institutions accounting for nearly 100 percent of volume on day two.
In prediction markets, the sector has grown from under $100 million in monthly volume in early 2024 to a weekly volume of $14.4 billion by June 2026, with major HFT firms including DRW, Susquehanna, and Jane Street building dedicated desks. Combined sector volume reached $47.5 billion in June 2026, driven by the World Cup as a primary catalyst. A new uniform-loss automated market making framework (Zhu, 2026, arXiv:2607.17428) addresses the unique liquidity provision challenges of binary-outcome prediction markets by designing pricing functions for which instantaneous loss-versus-rebalancing is proportional to pool value and independent of token price, extending to dynamic liquidity management that implements prescribed cumulative loss schedules. The CFTC proposed a comprehensive new regulatory framework on June 10, 2026, replacing blanket prohibitions on event contracts with a contract-by-contract public interest review framework, including the first formal definition of "gaming." Nine states are in active litigation with the CFTC over prediction market regulation, with Minnesota enacting the nation's first outright ban. Portnaya (2026) documents that systematic pricing wedges of 5.6 to 11 percentage points persist between Polymarket and exchange-traded option benchmarks for identical Bitcoin threshold contracts. The convergence between prediction markets and traditional options infrastructure has accelerated: Cboe launched its Cboe Predicts suite on June 23, 2026, featuring XSP binary options cleared through the OCC, and Charles Schwab partnered with Cboe to offer binary S&P 500 options. FanDuel and CME Group launched FanDuel Predicts on December 22, 2025, in five U.S. states, offering binary Yes/No event contracts priced at $0.01–$0.99 and settled against CME benchmarks, with a phased national rollout through early 2026.[^c6] Cboe also received SEC approval for extended trading hours for select single-stock options, launching July 13 with pre-market 7:30 a.m. ET and post-market to 4:15 p.m. ET windows. In Europe, ESMA classified prediction market event contracts as binary options under MiFID II in July 2026, subjecting them to the existing EU ban on retail binary options.
Crypto options have seen Bitcoin options open interest surpass futures for the first time, the launch of CME Bitcoin Volatility futures (June 1, 2026), and the emergence of a regulated 24/7 crypto derivatives trading framework — CME Group launched 24/7 cryptocurrency futures and options on May 29, 2026, generating $50 million in opening weekend volume, with Robinhood Markets, Ripple Prime, and Wedbush Securities among the first platforms to support the extended schedule. Deribit's Bitcoin options open interest climbed back to $31.3 billion on May 21, 2026, overtaking BlackRock's IBIT at $27 billion after IBIT had briefly surpassed Deribit in April for the first time since ETF options launched in November 2024; ahead of the June 26 monthly expiry carrying approximately $10.6 billion in notional open interest, Deribit's DVOL index sat at roughly 42 percent. By mid-July 2026, BTC options sentiment turned positive with the put/call ratio dropping to 0.59 (a six-month low) and DVOL declining from 48 to 40. Exchange infrastructure continues to evolve: Cboe added daily DJX expirations and extended single-stock options trading hours, and MEMX received regulatory approval for MX2 Options.
On June 11, 2026, the SEC proposed rescinding Regulation NMS Rule 611 and 610(e), the most significant equity market structure reform in decades. The proposal eliminates the trade-through prohibition and locked/crossed market restrictions, aiming to simplify market structure and reduce costs.
SpaceX completed the largest IPO in history on June 12, 2026, listing on the Nasdaq under ticker SPCX. The stock surged 67 percent to $225.64 within three trading sessions before retracing to $150 within two weeks, driven by a tiny public float of 4.24 percent and a staggered 15-stage lockup. The S-1 filing disclosed cumulative losses of more than $37 billion since the company's founding and a dual-class share structure that leaves founder Elon Musk with majority voting control. The options debut on June 17 set single-stock records with 1.8 million contracts traded. Ahead of the options debut, a ZeroHedge thesis summarized in an HTX market analysis argued that the combination of a low free float, concentrated retail call-buying, and the options listing could trigger a gamma squeeze toward an extreme $400 share price.[^c5]
CME Group launched Bitcoin Volatility Index futures on June 1, 2026, the first CFTC-regulated pure-volatility crypto derivative, settling to the CME CF Bitcoin Volatility Index (BVX) derived from real-time Bitcoin options order books.
In China, SSE ETF options traded 1.275 billion contracts (12.75亿张) in 2025 with 47.06 trillion RMB face value, while CFFEX-listed index options (IO, MO, HO) offer European-style cash-settled contracts with RMB 100 per index point multiplier. The market's first products — SSE 50 ETF options — launched on February 9, 2015, with 40 contracts, expanding to CSI 300 ETF options and CSI 300 index options on December 23, 2019.[^c4] The average realized 10-day volatility of the SSE 50 ETF was 22.59 percent, more than double the S&P 500's comparable figure. Yue, Zhang, and Tan (2020) document that 61.79 percent of call and 63.25 percent of put delta-hedged gains were negative, rejecting one-dimensional diffusion models and indicating a dominant volatility risk premium.
In closing auctions, Bender, Clapham, and Schwemmlein (2023) document that European markets with closing auction shares exceeding one third of daily volume experience 23 percent systematic overnight reversal of auction returns and reduced intraday liquidity, while Kandel, Rindi, and Bosetti (2008) provide foundational causal evidence that introducing a closing auction eliminates order-splitting incentives in the final minutes of continuous trading. SEC Commissioner Mark T. Uyeda questioned at the April 2026 Options Market Structure Roundtable whether legacy market maker entitlements and floor auction rules remain appropriate in the current 15-exchange options market.
Research Frontiers
The academic and quantitative research frontier continues to expand across multiple dimensions. Financial epiplexity (Noguer I Alonso, 2026) introduces a time-bounded minimum description length measure of learnable market structure, modeling alpha decay as bit migration and crowding as mutual compressibility. In volatility forecasting, the first systematic comparison of nine zero-shot time series foundation models against econometric benchmarks across 50 assets shows that only Tiny Time Mixers beats the Log-HAR benchmark and by only 1.3 to 1.8 percent, while choosing the right architecture matters more than the foundation-versus-econometric choice. A new decomposition of robustness in HFT market making identifies two economically meaningful dimensions — uncertainty tolerance and action robustness — showing that action robustness has substantially larger impact. In option-implied information, a shallow feedforward network with a single hidden layer effectively approximates implied density and implied volatility when arbitrage constraints are incorporated through a differentiable corrector, and an arbitrage-consistent neural SDE has been applied for the first time to intraday, multi-strike, multi-maturity crypto options data. Bali, Kelly, and Mörke (2026) construct a stock-level volatility disagreement measure (VDIS) from machine-learning realized-variance forecasts and find that a VDIS-sorted long-short delta-hedged straddle earns −5.14 percent per month, with evidence more consistent with mispricing than risk compensation. Guijarro-Ordonez, Pelger, and Zanotti (2025, Management Science) construct arbitrage portfolios as residual portfolios from conditional latent asset pricing factors, extracting time-series signals with a convolutional transformer to achieve significant out-of-sample Sharpe ratios on daily US equities.
Andersen, Todorov, and Zhou (2025) develop GLR-CUSUM sequential detectors for real-time identification of local Itô semimartingale violations in high-frequency data, identifying over 1,000 violation episodes across 3,500 S&P 500 futures trading days, with most lasting under 10 minutes. In behavioral option pricing, Shvimer and Herbon (2026) develop a two-stage behavioral model integrating subjective beliefs and execution propensity, reducing pricing errors by more than 30 percent versus Black-Scholes and Heston on 242,697 S&P 500 calls. Li and Papanicolaou (2026) demonstrate that deep neural networks achieve superior statistical arbitrage performance when stocks are indexed by market-capitalisation rank rather than by company name, due to enhanced mean-reverting properties of residual returns in rank space.
New leverage propagation SV models (STAR-SV, HARA, LPSV) capture magnitude-dependent and multiscale leverage transmission with closed-form propagation functions, tested on Bitcoin, Nasdaq, and S&P 500 data. The foundational measure v(s) = Var[y_t | y_{t-s} < 0] − Var[y_t] was introduced by Catania (2022) to quantify leverage propagation. The Itô-Wentzell formula and Dupire stochastic PDE for LSV models (Lucic, 2026) derive a conditional forward equation with a Rao-Blackwell estimator for the leverage function. A MACD portfolio optimization framework (Eccles, 2026) provides a mathematical foundation for MACD-type trading signals as filtered estimates of latent drift in a partial-information stochastic control problem, while Gruszka and Szwabiński (2020) confirm through Heston-model Monte Carlo that MACD strategies outperform passive benchmarks only when drift is large.
In portfolio optimization, a decision-focused learning framework with smooth top-k operators addresses the NP-hard cardinality-constrained tangent portfolio problem, tested across four major equity markets.[^c3] A Cholesky-based asset selection heuristic (OSCAR) achieves within 1 percent of optimal CPLEX solutions in seconds. Acciaio, Marini, and Pammer (2025) prove linear convergence and existence conditions for Bass local volatility model calibration. In market microstructure, a new equilibrium characterization of limit order markets with uncertain informed trading participation derives power-law price impact whose exponent depends on both the asset value tail and the full distribution of informed trader count. An explicit arbitrage-free construction of risk-neutral marginals (Qin et al., 2026) guarantees butterfly and calendar arbitrage freedom with closed-form densities. An end-to-end parametric portfolio policy using a transformer achieves 0.55 Sharpe on 16 CME futures with 0.02 daily turnover.
An exact characterization of adapted law invariance establishes equivalence to a recursive one-step conditional-law representation for time-consistent dynamic risk measures, complementing the Kupper-Schachermayer rigidity theorem which shows that terminal-law invariance forces entropic risk measures. The Barron-type universal approximation theorem for Bayesian neural SDE calibration provides robust bounds on the implied volatility surface via Langevin-type sampling. In execution, a DQN agent reduces implementation shortfall under dynamic fees in a DEX simulator by 13.3 basis points, while a signature-based optimal execution framework reduces path-dependent execution to a concave quadratic programme. A physics-informed Koopman-PINN framework (Zhu et al., 2026, AAAI) achieves accurate five-parameter Heston model recovery from high-frequency observations. A unified optimal adaptive market making framework (Zeng, 2026) extends Avellaneda-Stoikov and Glosten-Milgrom paradigms to perpetual futures, deriving a PnL decomposition theorem, Master APY Formula, and Kelly-optimal leverage with ruin boundaries.