Paper: arXiv 2408.01642
Authors: Jimin Lin, Guixin Liu
Abstract
The additive process generalizes the Lévy process by relaxing its assumption of time-homogeneous increments and hence covers a larger family of stochastic processes. Recent research in option pricing shows that modeling the underlying log price with an additive process has advantages in easier construction of the risk-neural measure, an explicit option pricing formula and characteristic function, and more flexibility to fit the implied volatility surface. Still, the challenge of calibrating an additive model arises from its time-dependent parameterization, for which one has to prescribe parametric functions for the term structure. For this, we propose the neural term structure model to utilize feedforward neural networks to represent the term structure, which alleviates the difficulty of designing parametric functions and thus attenuates the misspecification risk. Numerical studies with S&P 500 option data are conducted to evaluate the performance of the neural term structure.
Complexity vs Empirical Score
- Math Complexity: 9.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper is mathematically dense, featuring advanced stochastic calculus, Lévy processes, infinite divisibility, and specific derivations like Lemma 2.1 and Proposition 2.3, placing it at the high end of complexity. It demonstrates strong empirical rigor through calibration with S&P 500 option data, comparison to parametric models, and analysis of robustness across historical surfaces, making it backtest-ready despite focusing on theoretical construction.
Research Flowchart
flowchart TD A["Research Goal: Assess Neural Term Structure for Additive Process Calibration"] --> B["Data: S&P 500 Options Data"] B --> C["Model: Neural Term Structure<br>Feedforward Neural Networks"] C --> D["Computation: Calibrate Model to Market Data"] D --> E["Output: Improved Fit to Implied Volatility Surface"] E --> F["Outcome: Reduced Misspecification Risk<br>Explicit Pricing Formula Achieved"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a neural term structure model that uses feedforward neural networks to represent the term structure of additive processes for option pricing.
Source sentence
we propose the neural term structure model to utilize feedforward neural networks to represent the term structure, which alleviates the difficulty of designing parametric functions and thus attenuates the misspecification risk.
Assumes 1
- µ : R+ →R and σ, α, β : R+ →R+ are continuous functions model
Source sentence
Assumption 2.2. µ : R+ →R and σ, α, β : R+ →R+ are continuous functions and we have
Data
- S&P 500 implied volatility surface options commercial
Source sentence
We choose the S&P 500 implied volatility surface on 2022-06-30, quoted from Bloomberg, which is described by options in the grid of 13 moneynesses from 0.6 to 1.75 and 43 maturities from one week to approximately two and a half years.
Costs
Not applicable
Finds 2
- The neural term structure can fit the synthetic surface generated by the parametric term structure and calibrates better to real S&P 500 market data than the chosen parametric term structure. pricing-calibration
Source sentence
Then with an S&P 500 implied volatility surface, we show that the neural term structure better calibrates to the real market data than the chosen parametric term structure.
- Mean‑square calibration error for the jointly calibrated neural term structure η(t,τ) is slightly higher (0.98, 0.57, 1.00 ×10⁻²) than the single‑surface version η(τ) (0.71, 0.43, 0.39 ×10⁻²), but provides smooth, robust term‑structure surfaces across dates. performance
Source sentence
τ). ... The jointly calibrated η(t, τ) has a slightly higher calibration error than the separately calibrated η(τ).
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 6 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (35)
- Financial modelling with jump processes (2004)
- Normal variance-mean mixtures and z distributions (1982)
- Handbook of the Logistic Distribution (1991)
- Approximation by superposition of sigmoidal functions (1989)
- Dynamics of implied volatility surfaces (2002)
- Recovering Volatility from Option Prices by Evolutionary Optimization (2005)
- Fast exact joint S&P 500/VIX smile calibration in discrete and continuous time (2024)
- Building arbitrage-free implied volatility: Sinkhorn’s algorithm and variants (2019) — link
and 27 more
- Bass construction with multi-marginals: Lightspeed computation in a new local volatility model (2021)
- Making Markov Martingales Meet Marginals: With Explicit Constructions (2002)
- Markov-komposition und eine anwendung auf martingale (1972)
- A note on sufficient conditions for no arbitrage (2005)
- The range of traded option prices (2007)
- The variance gamma process and option pricing (1998)
- Normal inverse Gaussian distributions and stochastic volatility modelling (1997)
- A jump-diffusion model for option pricing (2002)
- Option pricing when underlying stock returns are discontinuous (1976)
- Jumps and stochastic volatility: Exchange rate processes implicit in deutsche mark options (1996)
- Managing smile risk (2002)
- Notes on option pricing I: Constant elasticity of diffusions, 1975 (1996)
- A closed-form solution for options with stochastic volatility with applications to bond and currency options (1993)
- Pricing with a smile (1994)
- The pricing of options and corporate liabilities (1973)
- Th 'e orie de la sp 'e culation (1900)
- An overview of gradient descent optimization algorithms (2016) — link
- Lecture 6.5-rmsprop: Divide the gradient by a running average of its recent magnitude (2012)
- Adam: A method for stochastic optimization (2014) — link
- TensorFlow: a system for large-scale machine learning (2016)
- Automatic differentiation in PyTorch (2017)
- L 'e vy processes and infinitely divisible distributions (1999)
- Characteristic Functions (1970)
- Families of distributions arising from distributions of order statistics (2004)
- A fast Monte Carlo scheme for additive processes and option pricing (2023)
- Explicit option pricing with additive processes (2023)
- Additive logistic processes in option pricing (2021)
0 of 35 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.