2022 Q4 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: 2022-10-01 to 2022-12-31
- Passed rule review: 9
- Sources: 6
Topic distribution¶
Domains¶
- volatility: 7
- lifecycle infrastructure: 1
Methods¶
- financial ml: 5
- research methods: 5
Facets¶
- instrument vix options: 3
- instrument index options: 1
- structure straddle: 1
Passed rule review¶
Nowcasting Stock Implied Volatility with Twitter¶
- Published: 2022-12-31
- Source: arXiv Quantitative Finance History
- Publication status:
preprint - Original source: Open original source
Why it matters¶
This work proposes a nowcasting pipeline using random forests and Twitter signals, which is directly relevant; without delay and cost-aware constraints, there is risk of over-claiming signal utility. (abstract:S1, abstract:S2, abstract:S3, abstract:S4, abstract:S6, abstract:S7)
Main author claims¶
- The authors predict next-day stock EOD implied volatility with random forests and evaluate on a universe of 165 liquid US stocks across 11 sectors. (
abstract:S1,abstract:S3) - They state ablation reveals added value from attention and sentiment features extracted from Twitter. (
abstract:S2) - They find sector differences in predictability; potential drivers include social-media attention or lower option liquidity, with value highest in lower-IV regimes but regime/sector dependent. (
abstract:S4,abstract:S5,abstract:S6,abstract:S7)
Data, method, or discussion scope¶
Evidence scope is an abstract summary of experiments across 11 sectors and a six-year OOS period; metric definitions, train/test splits, model hyperparameters, and delay controls are not included. (abstract:S3, abstract:S4, abstract:S6, abstract:S7)
Main limitations¶
No publication-timing, text-noise filtering, or Twitter-lagging protocol is provided, limiting causal direction assessment in operational time windows. (abstract:S2, abstract:S3, abstract:S6)
Forecasting swap rate volatility with information from swaptions¶
- Published: 2022-12-29
- Source: Journal of Futures Markets
- Publication status:
peer_reviewed - Original source: Open original source
Why it matters¶
If robust, out-of-sample predictive power of model-free implied swap volatility across regimes is directly relevant for fixed-income risk workflows; here it remains abstract-level and needs fuller specification. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)
Main author claims¶
- The authors investigate and claim that model-free implied volatility predicts future realized volatility of swap rates. (
abstract:S1,abstract:S2) - They claim model-free implied volatility has superior predictability versus lagged realized volatility and GARCH-type conditional volatility. (
abstract:S3) - They further claim this superiority holds out of sample, across different market states, and for longer horizons. (
abstract:S4)
Data, method, or discussion scope¶
Scope includes prediction tasks, tenor cross-sections, and out-of-sample claims; it lacks loss functions, error distributions, estimation frequency, and execution-oriented metrics. (abstract:S1, abstract:S2, abstract:S4)
Main limitations¶
Relative-performance wording lacks complete competitor definitions and explicit significance or multiple-testing controls. (abstract:S3, abstract:S4)
Beyond Surrogate Modeling: Learning the Local Volatility Via Shape Constraints¶
- Published: 2022-12-20
- Source: arXiv Quantitative Finance History
- Publication status:
preprint - Original source: Open original source
Why it matters¶
Comparing GP and NN approaches for no-arbitrage interpolation affects local-vol surface construction stability; statements about GP out-of-sample performance and NN smoothness require metric-specific interpretation. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)
Main author claims¶
- The authors evaluate two ML methods for no-arbitrage interpolation: a finite-dimensional GP regression with no-arbitrage constraints and an NN with arbitrage penalties on implied volatilities. (
abstract:S1) - They claim the GP approach is arbitrage-free by construction, whereas arbitrage is only penalized under SSVI and NN. (
abstract:S3) - They claim GP yields best out-of-sample calibration error while NN provides smoother local volatility and better backtesting performance. (
abstract:S4)
Data, method, or discussion scope¶
Evidence scope is abstract-level method comparison and directional performance statements only, without explicit metrics, training windows, or segmentation tests. (abstract:S1, abstract:S2, abstract:S4)
Main limitations¶
Claims versus SSVI and better backtesting are not tied to explicit reproducible metrics, and robustness under low liquidity or jump regimes is not specified. (abstract:S2, abstract:S3, abstract:S4)
Deep learning and American options via free boundary framework¶
- Published: 2022-12-09
- Source: arXiv Quantitative Finance History
- Publication status:
preprint - Original source: Open original source
Why it matters¶
The method reformulates American option free-boundary problems with a dual solution structure that is promising for numerically intensive workflows; however, claims of being an efficient alternative require caution without implementation detail. (abstract:S1, abstract:S2, abstract:S3, abstract:S8, abstract:S9)
Main author claims¶
- The authors propose using a Landau transformation to extract the early exercise boundary from the proposed deep learning framework. (
abstract:S1,abstract:S2) - They develop an auxiliary function combining DNN output with boundary behavior constraints and derive equations to approximate boundary and boundary derivatives directly. (
abstract:S4,abstract:S5,abstract:S6) - They claim efficiency and alternative-method performance when compared with existing numerical methods. (
abstract:S8,abstract:S9)
Data, method, or discussion scope¶
Scope is confined to methodological architecture and high-level comparison intent; no numerical error benchmarks, complexity bounds, or convergence rates are provided. (abstract:S1, abstract:S3, abstract:S8, abstract:S9)
Main limitations¶
Boundary values are approximated and efficiency is unquantified, limiting assessment of robustness for low-liquidity contracts or long-tenor structures. (abstract:S5, abstract:S6, abstract:S9, abstract:S8)
Understanding stock market instability via graph auto-encoders¶
- Published: 2022-12-09
- Source: arXiv Quantitative Finance History
- Publication status:
preprint - Original source: Open original source
Why it matters¶
The paper proposes a GAE reconstruction-error indicator from market-graph dynamics as a volatility proxy, useful as a signal concept; but at the abstract level this remains directional correlation evidence. (abstract:S1, abstract:S2, abstract:S3, abstract:S4, abstract:S5, abstract:S6)
Main author claims¶
- The authors represent co-movements as a correlation network and argue structure changes are central to understanding co-movement breakdowns. (
abstract:S2,abstract:S3) - They propose GAE edge reconstruction error as an indicator of spatial homogeneity changes and use it as a volatility proxy. (
abstract:S4) - The authors report: Using S&P 500 data (2015-2022), they report higher reconstruction error is associated with higher volatility and that out-of-sample autoregressive volatility modeling improves. (
abstract:S5,abstract:S6)
Data, method, or discussion scope¶
Evidence covers the proposed indicator and one period of S&P 500 experiments; it omits graph construction windows, threshold choices, train/test splits, and baselines. (abstract:S4, abstract:S5, abstract:S6)
Main limitations¶
No out-of-market generalization tests, and no mapping from signal latency/noise to execution frictions and trading decisions is given. (abstract:S4, abstract:S5, abstract:S6)
Why hasn’t “VOL” done better amidst stock market losses in 2022 with Logica’s Wayne Himelsein¶
- Published: 2022-12-08
- Source: The Derivative by RCM Alternatives
- Publication status:
unknown - Original source: Open original source
Why it matters¶
The description frames long-vol performance and a “new vol regime” discussion around 2022, which is useful for sentiment context; it is still interview framing, not an evidence-grade study. (description:S1, description:S2, description:S4, description:S5)
Main author claims¶
- The authors report: The episode claims the usual volatility spikes in down markets did not materialize to expected levels in 2022. (
description:S1,description:S2) - The authors report: It previews topics including weak vol sensitivity, contagion risk, feedback loops, and Logica’s strategy elements. (
description:S4,description:S6) - The authors report: The episode states it is informational only, prohibits specific trade recommendations, and includes investor risk warnings. (
description:S7,description:S9,description:S10,description:S11)
Data, method, or discussion scope¶
Evidence is limited to episode synopsis and chapter topics; no verifiable statistical sample, benchmark comparison, or trade records are provided. (description:S4, description:S6, description:S1, description:S7)
Main limitations¶
This is a promotional discussion outline rather than an empirical report; chapter headings do not map to a reproducible quantitative claim set. (description:S4, description:S6, description:S5)
Deep Weighted Monte Carlo: A hybrid option pricing framework using neural networks¶
- Published: 2022-12-08
- Source: arXiv Quantitative Finance History
- Publication status:
preprint - Original source: Open original source
Why it matters¶
This framework combines VAE-vol-surface compression with weighted paths to price vanilla and exotic options, which is deployment-relevant; yet “dynamic pricing” and “relative value signals” are asserted without explicit stability and error-budget details. (abstract:S1, abstract:S4, abstract:S5, abstract:S6, abstract:S7)
Main author claims¶
- They claim existing VAE approaches lack explicit underlying-asset dynamics and propose weighted Monte Carlo path weighting to correct this gap. (
abstract:S2,abstract:S3,abstract:S4) - They propose a neural network that assigns weights from latent space and combine VAE encoder with a weight assigner to denoise surfaces and price vanilla plus exotic options. (
abstract:S5,abstract:S6) - They claim the method can provide relative-value signals for option traders. (
abstract:S7)
Data, method, or discussion scope¶
Scope is limited to abstract claims about method design and intended outputs; sample details, loss functions, baselines, and latency/error controls are not included. (abstract:S1, abstract:S4, abstract:S5, abstract:S6)
Main limitations¶
At abstract level, it cannot support a trading-readiness claim; training stability, overfitting control, and execution boundaries (quotes/slippage) are absent. (abstract:S3, abstract:S5, abstract:S6, abstract:S7)
The Price of Higher Order Catastrophe Insurance: The Case of VIX Options¶
- Published: 2022-10-18
- Source: The Journal of Finance
- Publication status:
peer_reviewed - Original source: Open original source
Why it matters¶
The equilibrium model links equity and VIX/SVX market moments through affine jump-diffusion dynamics, affecting interpretation of higher-order risk pricing; however, calibration sufficiency and fit quality need full-paper validation beyond the abstract. (abstract:S1, abstract:S2, abstract:S3)
Main author claims¶
- The authors claim to build a tractable equilibrium pricing model with affine jump-diffusive state dynamics and representative agents with recursive preferences to jointly explain equity, VIX futures, SPX options, and VIX options data. (
abstract:S1) - They claim the calibrated model reproduces VIX futures returns, average implied volatilities in SPX and VIX options, and first and higher moments of VIX option returns. (
abstract:S2) - They claim time variation in implied-volatility shape and time-varying hedging relation between VIX and SPX options is captured by the model. (
abstract:S3)
Data, method, or discussion scope¶
The evidence is restricted to abstract-level calibration coverage claims without estimation constraints, pricing-error decomposition, or robustness checks included. (abstract:S1, abstract:S2, abstract:S3)
Main limitations¶
The abstract is theory-centric and omits parameter-update frequency, numerical stability, and market microstructure frictions required for operational use. (abstract:S1, abstract:S2, abstract:S3)
Time-Varying Skew in VIX Derivatives Pricing¶
- Published: 2022-10
- Source: Management Science
- Publication status:
peer_reviewed - Original source: Open original source
Why it matters¶
The abstract claims improved VIX derivative pricing with explicit in-sample and out-of-sample gains, which matters for model selection; however, “significantly outperforms” and uplift numbers need fuller implementation detail before deployment decisions. (abstract:S1, abstract:S3, abstract:S4, abstract:S2)
Main author claims¶
- The authors propose a reduced-form model with an independent stochastic jump intensity factor, co-jumps in VIX level and variance, and time-varying VIX variance mean. (
abstract:S1) - They fit the model to daily futures and option prices from 2007-04 to 2017-12 and claim 21.6% in-sample and 31.2% out-of-sample improvement versus nested competitors. (
abstract:S2,abstract:S3) - They claim more accurate tail behavior for VIX risk-neutral distributions at short and long maturities via time-varying skew largely independent of the VIX smile level. (
abstract:S4)
Data, method, or discussion scope¶
Scope includes model design, sample window, and reported uplift, but not the explicit loss metric definitions, competitor list, or confidence/uncertainty reporting. (abstract:S1, abstract:S2, abstract:S3, abstract:S4)
Main limitations¶
Abstract does not provide identification details for skew/jump parameters or how improved tail-fit translates into cost-feasible pricing and execution workflows. (abstract:S1, abstract:S4, abstract:S3)