Skip to content

2021 Q1 Quarterly Research Archive

Records scoring at least 40 within the primary scope pass rule review and are published without additional manual review. This page does not validate author claims or provide investment advice.

  • Coverage: 2021-01-01 to 2021-03-31
  • Passed rule review: 8
  • Sources: 5

Topic distribution

Domains

  • volatility: 6
  • hedging exposure risk: 3
  • microstructure: 3
  • execution costs: 3
  • option returns: 1

Methods

  • financial ml: 2
  • research methods: 2

Facets

  • instrument vix options: 3
  • instrument index options: 2
  • instrument single stock options: 1
  • horizon short dated: 1
  • structure straddle: 1

Passed rule review

Deep Hedging of Derivatives Using Reinforcement Learning

  • Published: 2021-03-29
  • Source: arXiv Quantitative Finance History
  • Publication status: preprint
  • Original source: Open original source

Why it matters

The study puts transaction costs, continuous state-action spaces, and alternative P&L objectives into an RL hedging framework, providing a cost-aware method case; real-market adaptability is not established. (abstract:S1, abstract:S7, abstract:S8, abstract:S9)

Main author claims

  • The authors report: Reinforcement learning can derive optimal hedging strategies with transaction costs, illustrated by showing the difference from delta hedging for a short call. (abstract:S1, abstract:S2)
  • The authors report: Using two Q-functions to track the expected value and expected squared value of cost expands the range of usable objective functions. (abstract:S7, abstract:S8)
  • The authors report: A hybrid approach combining accounting P&L and cash flow methods works well, and the valuation model need not correspond to the assumed price process. (abstract:S11, abstract:S12)

Data, method, or discussion scope

The paper considers a short call option under geometric Brownian motion and stochastic volatility processes, comparing delta hedging with optimal RL hedging. It explores accounting P&L and cash flow accounting methods. Specific numerical parameters or data are not provided in the abstract. (abstract:S4, abstract:S5, abstract:S10)

Main limitations

The abstract lacks details on algorithm convergence, computational cost, hyperparameter sensitivity, and out-of-sample performance. The effectiveness of the hybrid approach may depend on the chosen simple valuation model. The study is limited to numerical simulations and does not involve real market data or diversified derivative books. (abstract:S11, abstract:S4)

Smile‐implied hedging with volatility risk

  • Published: 2021-03-26
  • Source: Journal of Futures Markets
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The authors add volatility-risk management to smile-implied hedging and report gains over benchmarks; without cost, frequency, and liquidity details, practical effectiveness is not established. (abstract:S3, abstract:S4)

Main author claims

  • The authors report: Smile-implied delta and delta-gamma hedging alone fail to achieve minimum variance under price-volatility correlation and underperform the BS benchmark. (abstract:S2)
  • The authors report: With added vega management, smile-implied delta-gamma-vega strategies outperform the BS approach and more sophisticated frameworks including stochastic volatility and jumps. (abstract:S3, abstract:S4)
  • The authors report: Large-scale empirical evidence from S&P 500 index options supports these conclusions. (abstract:S4)

Data, method, or discussion scope

The analysis uses large-scale S&P 500 index option data, comparing smile-implied delta, delta-gamma, and delta-gamma-vega strategies against BS and other complex models. The abstract does not provide the exact sample period or specific hedging error metrics. (abstract:S4)

Main limitations

The abstract does not report transaction costs, rebalancing frequency, or performance during market dislocations. Vega hedging requires trading additional volatility instruments, the liquidity and cost of which are not considered. Findings are limited to S&P 500 index options. (abstract:S3, abstract:S4)

Pricing VIX options with realized volatility

  • Published: 2021-03-13
  • Source: Journal of Futures Markets
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The study incorporates realized volatility into VIX option pricing and reports error improvements for GARV and Realized GARCH over return-only models; superiority as a valuation tool still requires full sample and error analysis. (abstract:S1, abstract:S4, abstract:S5)

Main author claims

  • The authors report: Models incorporating realized volatility significantly outperform competing daily-return-based models both in-sample and out-of-sample. (abstract:S4)
  • The authors report: A closed-form pricing formula is derived for the affine GARV model, and a novel approximation method is introduced for the nonaffine Realized GARCH model. (abstract:S2, abstract:S3)
  • The authors report: The Realized GARCH model offers the best pricing performance due to its fewer constraints and more flexible structure. (abstract:S5)

Data, method, or discussion scope

The empirical analysis uses VIX options data, comparing the GARV and Realized GARCH models against daily-return-based models in- and out-of-sample. Exact data period and option filtering criteria are not stated in the abstract. (abstract:S4, abstract:S1)

Main limitations

The abstract lacks details on sample period, option contract characteristics, and estimation methodology. The approximation for the Realized GARCH model may introduce bias, but approximation error is not discussed. The study is limited to VIX options, leaving applicability to other volatility products uncertain. (abstract:S4, abstract:S3)

Index Option Trading Activity and Market Returns

  • Published: 2021-03
  • Source: Management Science
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The authors report a conditional predictive relation between index-put order flow and subsequent weekly S&P 500 returns, making it candidate evidence on options-flow information; the abstract does not establish causality or timing effectiveness. (abstract:S2, abstract:S3)

Main author claims

  • The authors report: Weekly ISE index put order flow positively and robustly predicts subsequent weekly S&P 500 returns. (abstract:S2)
  • The authors report: The predictive effect is driven by net put buying and is stronger in high VIX periods and after macroeconomic announcements. (abstract:S3)
  • The authors report: Option-based risk protection strategies by retail investors best explain the findings. (abstract:S4, abstract:S5)

Data, method, or discussion scope

The study uses weekly ISE index put order flow and S&P 500 index returns, focusing on net put buying, VIX regimes, and macroeconomic announcement windows. It explores several rationales, including sentiment, market-maker information, and retail hedging. (abstract:S2, abstract:S3, abstract:S4)

Main limitations

The abstract does not disclose the sample period, full set of controls, or robustness checks. The favored explanation is the one that best fits the data, not a causally tested conclusion. The analysis is limited to a single exchange and equity index, with uncertain generalizability. (abstract:S2, abstract:S4, abstract:S5)

Volatility Estimation and Forecasts Based on Price Durations

  • Published: 2021-03-01
  • Source: Journal of Financial Econometrics
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The study revisits price-duration variance estimators and reports advantages in some simulation and forecasting settings, offering an additional high-frequency volatility tool whose performance depends on threshold, noise, and jump handling. (abstract:S1, abstract:S3, abstract:S6)

Main author claims

  • The authors claim that price duration estimators can be used for estimation and forecasting of integrated variance, and are affected by discrete spacing, microstructure noise, and finite jumps. (abstract:S3)
  • The paper develops asymptotic results for its nonparametric estimator under both idealized and bid–ask/time-discrete settings. (abstract:S4)
  • The authors claim that price duration estimators can extract more relevant information and produce more accurate forecasts than competing realized volatility and option-implied variance estimators, both in isolation and in forecast combinations. (abstract:S6)

Data, method, or discussion scope

The study uses simulation and empirical forecasting with high-frequency data, comparing price duration estimators against realized volatility and implied variance estimators. (abstract:S5, abstract:S6)

Main limitations

The estimators may be sensitive to the choice of threshold parameter; practical application may be affected by data availability and computational efficiency; performance during extreme volatility periods is not discussed. (abstract:S5)

The VIX index under scrutiny of machine learning techniques and neural networks

  • Published: 2021-02-03
  • Source: arXiv Quantitative Finance History
  • Publication status: preprint
  • Original source: Open original source

Why it matters

This preprint studies whether a small option subset and ML can approximate VIX, which is relevant to index replicability; the abstract does not establish executable arbitrage or a manipulation finding. (abstract:S7, abstract:S8, abstract:S9)

Main author claims

  • The paper says its models can approximate VIX using a reduced option subset. (abstract:S7)
  • The authors claim that by replicating the VIX, potential arbitrage opportunities between the VIX index and its derivatives can be exploited. (abstract:S8)
  • The authors claim that the results can give guidance to US regulators and CBOE investigating manipulation claims. (abstract:S9)

Data, method, or discussion scope

The study uses SPX options data and various methods including subset selection, Random Forests, SVMs, feed-forward, and LSTM neural networks to replicate the VIX. (abstract:S6, abstract:S7)

Main limitations

Replication may overlook specific technical details of the CBOE methodology; actual profitability of arbitrage opportunities is not verified; it is a preprint without peer review. (abstract:S4, abstract:S8, abstract:S9)

Information Content of Aggregate Implied Volatility Spread

  • Published: 2021-02
  • Source: Management Science
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The authors report incremental predictive relations between aggregate IVS, market returns, and macro news, making it a candidate forecasting signal; common informed trading is a proposed mechanism, not an identified causal conclusion. (abstract:S1, abstract:S2, abstract:S3, abstract:S5)

Main author claims

  • The paper reports a positive predictive relation between aggregate IVS and later equity-index returns over horizons from days to several months. (abstract:S1)
  • The authors claim that IVS's predictive power is incremental to existing predictors and can forecast macroeconomic news up to one year ahead. (abstract:S2, abstract:S3)
  • The authors claim that common informed trading in equity options offers an integrated explanation for IVS's ability to predict both stock returns and real economic activity. (abstract:S5)

Data, method, or discussion scope

The study uses US equity options data to construct aggregate IVS, performs in-sample and out-of-sample predictive tests, and analyzes its predictive power for macroeconomic announcements. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)

Main limitations

Controls for data-snooping bias are not described; applicability outside the US market is unknown; the specific mechanism of the informed trading channel is not fully elaborated in the abstract. (abstract:S5)

Maturity Driven Mispricing of Options

  • Published: 2021-01-19
  • Source: Journal of Financial and Quantitative Analysis
  • Publication status: peer_reviewed
  • Original source: Open original source

Why it matters

The authors report short-option return differences associated with expiry-week structure and propose an investor-attention explanation; the result is relevant to calendar-effect research, but the abstract cannot rule out liquidity or microstructure alternatives. (abstract:S1, abstract:S3)

Main author claims

  • The paper reports lower short-dated option returns when the expiry calendar contains four weeks rather than five. (abstract:S1)
  • The authors claim that the average return differential ranges from 16 to 29 bps per week for delta-hedged portfolios, and from 101 to 187 bps per week for straddles. (abstract:S2)
  • The authors claim that evidence based on earnings announcements and institutional holdings suggests investor inattention to exact expiration date, rather than underlying risk or transaction costs, explains the mispricing. (abstract:S3)

Data, method, or discussion scope

The study uses option market data from 1996-2017, analyzing returns of delta-hedged and straddle portfolios, and supporting a behavioral explanation with earnings announcement and institutional holding data. (abstract:S2, abstract:S3)

Main limitations

Other non-behavioral factors (e.g., tax or settlement mechanics) may not be fully ruled out; out-of-sample period performance is not mentioned; market maker adjustments may diminish future arbitrage opportunities. (abstract:S3, abstract:S4)