Papers
Topics
Authors
Recent
Search
2000 character limit reached

Stochastic Volatility, Jumps, and Rates: A Unified Framework for Option Pricing and Term-Structure Simulation

Published 27 May 2026 in q-fin.PM, q-fin.CP, q-fin.MF, q-fin.PR, and q-fin.ST | (2605.27945v1)

Abstract: This study develops an integrated stochastic modeling framework for pricing short and medium-maturity equity options and assessing interest-rate risk using the Heston (1993), Bates (1996), and CIR (1985) models. We calibrate the Heston model using both the Lewis (2001) Fourier inversion and the Carr-Madan (1999) FFT approach, finding near-identical parameter sets, which is consistent with the calibration stability reported in recent studies such as Agazzotti et al. (2025). Extending the model to Bates shows that jump intensities converge to values effectively equal to zero for 60-day maturities, echoing empirical findings that jumps contribute marginally to short-term smile fitting. We further compare our calibration approach with the joint volatility-surface and variance-term-structure framework proposed by Yoo (2025), confirming that standard Heston/Bates calibration remains robust for the maturities considered. Finally, we calibrate the CIR short-rate model to the Euribor term structure, generating positive and economically consistent forward-rate scenarios in line with recent stochastic-rate option-pricing research by Jeon and Kim (2025). Overall, our results show that continuous stochastic volatility dominates near-term pricing dynamics, while stochastic interest rates materially influence valuations beyond one year.

Summary

  • The paper presents a unified modeling framework combining stochastic volatility (Heston), jump-diffusion (Bates), and stochastic interest rates (CIR) to enhance option pricing.
  • Empirical calibration using Fourier inversion and FFT methods yields robust parameter estimates with minimal pricing errors across different maturities.
  • Monte Carlo simulations reveal that stochastic rate dynamics significantly affect medium- to long-term derivative valuations and risk management.

Unified Stochastic Modeling Framework for Derivative Pricing

Overview

The paper "Stochastic Volatility, Jumps, and Rates: A Unified Framework for Option Pricing and Term-Structure Simulation" (2605.27945) develops a comprehensive framework for modeling and pricing short- and medium-maturity equity options while simultaneously capturing interest-rate risk. The approach integrates the Heston stochastic volatility model, its jump-diffusion extension via Bates, and the CIR affine model for short rates. Critical calibration techniques—Lewis Fourier inversion and Carr-Madan FFT—are employed, and their numerical robustness is validated. The study’s empirical calibration leverages market data from SM Energy options and the Euribor curve, offering a systematic perspective on the interplay between volatility, jump risk, and stochastic rates across maturities.

Theoretical Formulation

Stochastic Volatility and Jump-Diffusion

The Heston model provides a tractable stochastic volatility structure, embedding mean-reversion and volatility clustering effects. The Bates model augments Heston with price jumps governed by Poisson arrivals and Normally distributed log-jump sizes, enabling the capture of sudden asset price dislocations commonly encountered in event-driven markets.

The model calibrations exploit the characteristic function, enabling semi-analytical pricing via Fourier methods (Lewis, 2001; Carr-Madan, 1999). Both the Lewis inversion and Carr-Madan FFT approaches return nearly identical parameter sets, confirming calibration stability and consistency in the estimation process. The volatility-of-volatility parameter systematically collapses towards zero across maturities, revealing that market-implied dynamics are nearly deterministic—this is corroborated by minimal pricing errors when using either calibration technique.

Stochastic Interest Rates

The CIR model introduces stochastic dynamics for the short-term interest rate, enforcing positivity and mean-reversion—characteristics essential for arbitrage-free pricing and term-structure modeling. Calibration to the Euribor term structure produces economically reasonable parameters: rapid mean reversion (K=2.0), equilibrium rate ≈4.2%, and moderate volatility. The Feller condition is satisfied, ensuring strictly positive rates.

Monte Carlo simulation under the calibrated CIR parameters generates realistic forward-rate scenarios, with a mean of 3.73% and a right-skewed distribution characteristic of square-root processes. The resulting discount factors are systematically lower than those implied by a flat-rate curve, confirming the material impact of stochastic rates on medium- to long-maturity derivative valuations.

Empirical Calibration and Pricing Results

Calibration exercises for SM Energy option data (15, 60, and 120 days) consistently find:

  • Stochastic volatility is adequate for short- and medium-maturity option pricing: Vol-of-vol parameter is virtually zero, and model errors are minimal. Correlation parameter is close to +1, deviating from typical equity leverage effect, but error analysis supports the empirical fit.
  • Jump intensity in Bates model converges to zero for maturities ≤60 days: Indicates negligible market-implied jump risk. Addition of jump components does not materially improve fit or alter pricing outputs, consistent with calm market regimes or well-captured skew by stochastic volatility alone.
  • Fourier-based calibration (Lewis vs Carr-Madan) returns near-identical results: Confirms numerical robustness and that model parameters are well-identified by the market data for the maturities studied.

Monte Carlo pricing for 20-day Asian call and 70-day 95% moneyness Asian put demonstrates high-pricing precision (standard errors <0.02), further validating the calibration quality.

CIR calibration yields a term-structure fit with RMSE ≈3.65 bps and MAE ≈2.25 bps. Monte Carlo analysis shows substantial distributional spread in 12M forward rates (5th percentile: 0.19%, 95th: 11.13%). Discount factors obtained from CIR simulations are more conservative than flat-curve pricing, with a mean reduction of ≈2.2%. This underscores the necessity of stochastic rate modeling for valuation accuracy beyond one year.

Implications and Model Limitations

Practical Implications

  • Short-maturity options: Continuous stochastic volatility captures market-implied dynamics; jump risk is not evidenced for the datasets and horizons considered.
  • Medium-maturity and longer-dated contracts: Stochastic interest rates produce significant valuation differences, especially in environments where yield curves are steep or volatile.
  • Calibration robustness: Stability of Fourier-based parameter estimation and tight pricing errors support deployment in production valuation pipelines.
  • Risk management: Monte Carlo path simulations for both volatility and rate models offer granular scenario analysis, capturing distributional tail risks relevant for PFE/PVaR calculations.

Model Limitations

  • Single-underlying calibration restricts generalizability; cross-sectional studies are required for broader applicability.
  • Bates jump structure assumes Normally distributed jumps, which may not reflect empirically observed heavy-tailed phenomena.
  • CIR positive-rate enforcement reduces applicability in negative-rate regimes.
  • Calibration uses mid-quotes, omitting bid-ask uncertainties or liquidity premiums.
  • Risk-neutral measure is assumed universally; alternative pricing kernels or incomplete-market effects are not explored.

Future Directions

The study outlines several technical enhancements:

  • Flexible jump structures (double-exponential, tempered stable) may capture skew/kurtosis in stressed or highly asymmetric markets.
  • Joint calibration of volatility surface and variance term structure may further stabilize parameters, especially across maturities.
  • Extension of CIR to multi-factor affine models could better capture curve shifts and structural rate changes.
  • Incorporation of stochastic correlation and rough volatility dynamics to address high-frequency features.
  • Expansion to multi-asset derivatives and portfolio-level risk modeling would test robustness in higher dimensions.

Conclusion

This research establishes a robust, integrated stochastic modeling approach for derivative pricing. Empirical calibration for SM Energy options and the Euribor term structure demonstrates that, for short and medium maturities, stochastic volatility alone suffices, and jump-diffusion extensions (Bates) activate only marginally. The CIR model’s realistic rate simulations reinforce the importance of stochastic rates for horizon-dependent valuation. The pipeline's flexibility and empirical grounding situate it as a strong candidate for contemporary option pricing and risk management, with opportunities for further theoretical and practical advancements in model generalization and calibration techniques.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Explain it Like I'm 14

What this paper is about (big picture)

This paper is about building a smart-but-practical way to price stock options and to simulate where interest rates might go. The authors combine three well-known models:

  • Heston: volatility (how much prices wiggle) can change over time
  • Bates: same as Heston but also allows sudden jumps in prices
  • CIR: a model for interest rates that keeps them positive and tends to pull them back toward a typical level

They test these models on real market data for short and medium timeframes (about 2 weeks to 4 months for options, up to 1 year for interest rates) to see which features matter most.

What questions the paper tries to answer

  • Do we really need “jumps” to explain prices of options that expire soon (15–120 days)?
  • If we use two different math tools to price options with the same model, do we get the same answers?
  • How much do changing interest rates matter for pricing, especially as we look further into the future?

How they did it (methods made simple)

Think of modeling like tuning a video game car so it behaves like a real car on a track. “Calibration” is the tuning step—adjusting the model’s knobs until its prices match market prices.

  • Heston model (changing volatility): Like weather that can be calm or stormy and slowly shifts over time.
  • Bates model (jumps): Adds occasional sudden “earthquakes” in prices on top of the usual weather.
  • CIR interest-rate model: Interest rates are like a springy ruler—pulled toward an average level and never going below zero.

How they priced options:

  • They used two different but related “math calculators” based on the same idea (Fourier transforms) to turn the model’s description into option prices:
    • Lewis inversion and
    • Carr–Madan FFT
    • Think of it like pressing the same recipe through two different kitchen machines—you expect the same soup at the end.

How they checked the results:

  • They fit the Heston and Bates models to market option prices for SM Energy at 15, 60, and 120 days.
  • They priced an Asian option (an option on the average price over time) using Monte Carlo simulation—like rolling many dice to see all possible outcomes and averaging them.
  • They fit the CIR model to the Euribor interest-rate curve using its built-in formulas, then simulated many possible future interest-rate paths to see the range of outcomes.

What they found and why it matters

Here are the main takeaways, with short explanations of why they’re useful:

  • For short-term options, changing volatility is enough.
    • Result: The Heston model fit the market well. The “volatility-of-volatility” (how much volatility itself wiggles) came out very close to zero for 15–120 days.
    • Why it matters: For options expiring soon, the market didn’t need complicated extra features to be explained.
  • Jumps didn’t help for 60-day options.
    • Result: In the Bates model, the “jump intensity” calibrated to essentially zero for 60-day options.
    • Why it matters: If sudden price shocks don’t improve the fit, you can use a simpler model (Heston) without losing accuracy.
  • Two different pricing tools gave almost the same parameters and prices.
    • Result: Lewis and Carr–Madan methods produced nearly identical Heston (and Bates) parameters and similar small errors.
    • Why it matters: This shows the calibration is stable and not sensitive to which numerical tool you use.
  • Asian option prices were stable and matched across methods.
    • Result: A 20-day Asian call had almost the same fair value under both Heston calibrations (difference less than a cent).
    • Why it matters: Extra confirmation that the modeling and numerical tools are consistent.
  • Interest rates: positive, mean-reverting, and important beyond one year.
    • Result: The CIR model matched the Euribor curve very closely. Simulations showed rates tend to move upward toward about 4%+ on average, with some uncertainty.
    • Why it matters: Even if short-term stock options don’t need fancy features, interest-rate movements become important as you look further out in time. Using a flat interest rate can overvalue future cash flows compared with a realistic, moving-rate model.

Why this research is useful (simple implications)

  • For near-term option pricing (weeks to a few months), a well-calibrated stochastic-volatility model like Heston is usually enough. You probably don’t need to add jumps unless the market is especially turbulent.
  • You can trust the results from standard Fourier-based pricing methods; they tend to agree when implemented carefully.
  • As maturities get longer, don’t ignore interest rates. A realistic rate model like CIR can change valuations by a few percent compared with assuming a flat rate.
  • Overall, the paper’s “unified framework” shows how to pick the right level of model complexity for the task: keep it simple for short maturities, and bring in stochastic interest rates for longer horizons.

A quick note on limits and future ideas

  • The study focuses on one stock and calm market conditions; stressed markets might need jumps or richer models.
  • The jump sizes assumed a normal shape; other shapes could fit extreme events better.
  • CIR keeps rates positive, which is good now but may not fit situations with negative rates.
  • Future work could try more flexible jump models, multi-factor interest-rate models, and newer “rough volatility” ideas to handle fast market moves.
View Paper Prompt  View All Prompts 

Knowledge Gaps

Unresolved knowledge gaps, limitations, and open questions

The following list summarizes what remains missing, uncertain, or unexplored in the study, phrased to guide concrete follow-up research:

  • External validity: Results are based on a single equity (SM Energy) and one market period; test across multiple assets, sectors, and stressed regimes (e.g., earnings, macro events) to assess generalizability of near-zero vol-of-vol and negligible jumps.
  • Maturity coverage: Equity option calibration only spans 15–120 trading days; investigate whether the “jumps ≈ 0” and “σ_v ≈ 0” findings persist for longer-dated options (e.g., 6–24 months) where term-structure effects and jump risk may be more pronounced.
  • Parameter identifiability and uncertainty: No confidence intervals, standard errors, or posterior distributions are reported; quantify identifiability (especially when σ_v → 0) and provide parameter uncertainty (e.g., via bootstrapping, Fisher information, or Bayesian inference).
  • Correlation instability: The Heston correlation switches from strongly positive (~+0.99) to near zero/negative across calibrations; analyze whether this reflects non-identifiability, numerical artifacts, or genuine asset-specific leverage features, and assess its pricing impact.
  • Loss function design: Calibration minimizes unweighted price MSE; evaluate alternative objectives (e.g., vega-weighted errors, implied-vol errors, robust Huber loss) and their impact on parameter stability and fit across strikes/maturities.
  • Market microstructure and liquidity: Mid-quotes are used without bid-ask weighting or liquidity filters; incorporate bid-ask spreads, exclude stale/deep-OTM quotes, and enforce no-arbitrage constraints to reduce noise-induced bias.
  • Dividends and carry: Put–call parity is applied assuming a flat risk-free rate (1.5%) and no dividends; verify and incorporate dividend yields (or funding/borrow costs) to avoid parity-induced bias for SM Energy.
  • Cross-currency/rate inconsistency: Equity options are priced with a flat 1.5% rate while the rate model is calibrated to EUR Euribor; align the rate environment (e.g., USD OIS for a US stock) or explicitly assess sensitivity to the risk-free proxy.
  • Joint equity–rate modeling: CIR is calibrated separately and not coupled with Heston/Bates for equity option pricing; build and test an integrated stochastic volatility–stochastic rates model (including equity–rate and variance–rate correlations) to quantify impacts beyond one year.
  • Hedging and out-of-sample validation: No tests of hedging performance or out-of-sample pricing stability are reported; run rolling calibrations and delta/vega hedging backtests to evaluate practical robustness.
  • Model selection for jumps: The conclusion “λ ≈ 0” is based on in-sample fit; conduct likelihood-based tests (AIC/BIC, likelihood-ratio tests) and out-of-sample comparisons to formally assess whether Bates adds explanatory power.
  • Event risk and time-varying jumps: Calibration does not consider event windows (earnings, M&A, macro releases) or state-dependent jump intensities; test time-inhomogeneous jump models or regime-switching intensities for event-driven risk.
  • Numerical sensitivity (Fourier methods): Choices such as damping (α), grid spacing, truncation (u_max=200), and FFT aliasing are not stress-tested; provide sensitivity analyses to ensure numerical stability is not driving parameter conclusions.
  • Heston path discretization for Asian options: The Monte Carlo setup for arithmetic Asians under Heston lacks details on variance discretization (e.g., QE/Broadie–Kaya, full truncation), bias control, and variance reduction (antithetic, control variates); quantify discretization bias and convergence.
  • Product definition consistency: The 70-day put section mixes a European formula with “Asian put” Monte Carlo results; clarify the product type and ensure pricing methodology matches the payoff definition.
  • CIR simulation scheme: Despite the availability of exact sampling for CIR, Euler discretization is used; assess bias from Euler (especially near the Feller boundary), compare to exact/noncentral chi-square sampling, and quantify effects on discount factors.
  • Curve construction and instruments: Euribor curve is spline-interpolated from short tenors without full market bootstrapping (e.g., deposits/FRAs/swaps) or arbitrage-free smoothing; implement a standard bootstrapping pipeline with correct day-counts and compounding conventions.
  • Rate model scope: CIR is fitted only up to 12 months; extend to multi-year horizons and test multi-factor affine models to capture slope/curvature dynamics and structural shifts in the yield curve.
  • Option-rate interaction beyond DF: The study shows discount factor dispersion but does not quantify impacts on equity option prices or Greeks for longer maturities; explicitly price >1y equity options under stochastic rates and compare vs flat-rate benchmarks.
  • Real-world vs risk-neutral dynamics: All analyses are under Q without discussing risk premia; explore P-dynamics estimation (e.g., from time series) and the mapping to Q (market prices of risk), or assess model implications for implied risk premia.
  • Parameter regularization and priors: No regularization is used despite near-degenerate σ_v; test constrained calibration (e.g., priors on κ, θ, σ_v, ρ) to stabilize identification and prevent economically implausible solutions.
  • Data transparency and reproducibility: Exact data date, dividend assumptions, code, seeds, optimizer settings, and parameter bounds are not disclosed; provide these artifacts to enable replication and robustness checks.
  • Sensitivity to initial guesses and optimizers: Results rely on Differential Evolution + L-BFGS-B; report sensitivity to initial seeds, alternative optimizers, and multi-start strategies to ensure convergence is not local-minima driven.
  • Greek and risk diagnostics: Pricing errors are reported, but no analysis of model-implied Greeks or risk exposures is provided; compare Heston/Bates vs Black–Scholes Greeks and quantify model risk in hedging.
  • Negative-rate environments: CIR’s positivity is a known limitation but no alternative (e.g., shifted-CIR, Gaussian affine, or HW++) is benchmarked; test models suitable for low/negative-rate regimes and assess pricing differences.
  • Smile structure diagnostics: The study asserts good fit but does not report implied-vol smiles/term structure or deep-OTM behavior; examine wing fits, convexity, and extrapolation quality across strikes and maturities.
View Paper Prompt  View All Prompts 

Practical Applications

Immediate Applications

The following items translate the paper’s findings into deployable use cases across sectors. Each bullet notes the sector, the potential tool/product/workflow, and key assumptions or dependencies.

  • Heston-only short-dated option pricing and quoting
    • Sector: Finance (trading, market making), Software
    • What: Use the Heston model calibrated via Lewis (2001) and/or Carr–Madan (1999) FFT to price and quote short-to-medium maturity equity options (15–120 days) without jumps.
    • Why: The paper shows near-identical calibrations across Fourier methods and negligible jump intensity at 60 days; vol-of-vol collapses to ~0, implying simple, stable dynamics.
    • Tool/workflow: “Dual-transform calibration” service that runs Lewis and FFT side-by-side, compares parameters/errors (MSE/MAE), and publishes quotes with a stability flag.
    • Assumptions/dependencies: Sufficient strike coverage; calm-to-normal market regime; risk-neutral pricing; mid-quotes; may not generalize to stressed markets or different underlyings.
  • Automated jump-switch rule in model selection
    • Sector: Finance (risk, trading), Software
    • What: Calibrate Bates, then automatically switch to Heston if estimated jump intensity λ≈0 and jump variance ≈0 at target maturities (e.g., 30–90 days).
    • Why: The paper’s calibrations repeatedly drive λ to ~0 at 60 days, showing jumps add little near term.
    • Tool/workflow: Model-governance rule with AIC/BIC-style penalty that deactivates jumps to reduce overfitting and runtime.
    • Assumptions/dependencies: Stability may be asset- and regime-specific; periodic revalidation needed.
  • Asian option pricing under calibrated Heston
    • Sector: Finance (structured products, corporate hedging), Software
    • What: Price short-dated arithmetic Asian options with Monte Carlo using Heston parameters; adopt convergence guidance (~100k–200k paths).
    • Why: The paper demonstrates robust Asian pricing with minor method sensitivity and clear convergence.
    • Tool/workflow: MC pricer module with auto-tuning of path count and error bars; fee overlay (e.g., 4%).
    • Assumptions/dependencies: Path discretization and antithetic/variance reduction choices; short maturities favored by near-deterministic variance.
  • Realistic discounting for 6–18 month valuations using CIR
    • Sector: Finance (treasury, XVA, ALM), Energy, Insurance
    • What: Replace flat-curve discounting with CIR path-dependent discount factors for valuations beyond ~1 year.
    • Why: The paper finds average 1Y discount factor ~2.2% lower than a flat curve, materially affecting PVs.
    • Tool/workflow: “DF engine” that generates pathwise DFs, aggregates PV adjustments, and compares to flat-curve benchmarks.
    • Assumptions/dependencies: CIR calibration quality (Feller condition, curve fit); potential limitations in negative-rate regimes.
  • Rate-scenario generator for near-term risk dashboards
    • Sector: Finance (risk), Corporate Treasury
    • What: Use calibrated CIR to produce weekly/quarterly scenarios and percentiles for short-rate paths and their impact on discount factors.
    • Why: Paper’s 12M distribution illustrates meaningful uncertainty and skew; supports decisions on hedging and liquidity buffers.
    • Tool/workflow: Dashboard with scenario fans, DF distributions, and PV-at-risk metrics.
    • Assumptions/dependencies: Current curve quality; rate positivity assumption; limited horizon realism for one-factor models.
  • Model validation and governance checklist
    • Sector: Finance (model risk), Audit/Accounting (IFRS 13), Academia
    • What: Implement a standardized back-to-back calibration via both Fourier methods, error diagnostics, Feller checks, and jump necessity tests.
    • Why: Paper demonstrates calibration robustness and a transparent validation pattern.
    • Tool/workflow: Validation notebook/templates producing parameter comparisons, pricing residuals, and stability plots.
    • Assumptions/dependencies: Good market data (bid-ask, liquidity); governance thresholds for acceptable MSE/MAE.
  • Cost-aware compute settings and SLAs for MC pricing
    • Sector: Finance (quant platforms), Software/Cloud
    • What: Use convergence evidence to set default path counts and timeouts; autoscale MC capacity based on target standard error.
    • Why: Paper shows stable Asian pricing beyond ~100k paths, enabling predictable latency/cost trade-offs.
    • Tool/workflow: MC job manager with error-to-cost curves; standardized CI tests for pricing reproducibility.
    • Assumptions/dependencies: Stationarity of volatility over short horizons; hardware variability.
  • Education and curriculum modules
    • Sector: Academia, Professional Training
    • What: Teaching materials covering Heston/Bates via transform methods and CIR with calibration and simulation; comparisons and pitfalls.
    • Why: Paper’s unified pipeline is didactic and practically relevant.
    • Tool/workflow: Labs in Python/Julia/Matlab, datasets, auto-graders for calibration stability.
    • Assumptions/dependencies: Access to option chains and term structures; licensing for data.
  • Treasury/hedging playbooks for corporates
    • Sector: Corporate Treasury (Energy, Industrials)
    • What: For short-dated equity or commodity-linked exposures, deploy Heston (no jumps) for option structuring and use CIR DFs for >1Y cash-flow PV sensitivity.
    • Why: Paper shows diffusion-only suffices near term; rates matter more beyond a year.
    • Tool/workflow: Hedging policy that selects model by tenor; PV uplift monitor comparing flat vs stochastic discounting.
    • Assumptions/dependencies: Proxying commodity vol with equity data must be justified; data mapping required.
  • Retail/wealth advisory guardrails
    • Sector: Wealth/Fintech
    • What: For near-term option strategies (covered calls, protective puts), warn users that jump premia may be small in calm markets; emphasize rate risk in longer-dated products.
    • Why: Paper evidences negligible 60-day jump intensity and meaningful rate impact beyond 1y.
    • Tool/workflow: In-app nudges and risk notes; “rate-sensitivity” badges for structured notes >12 months.
    • Assumptions/dependencies: Suitability and disclosures; simplified messaging for non-expert users.

Long-Term Applications

These items require further research, scaling, or development before broad deployment. Each bullet notes likely sectors, possible tools/products, and key assumptions.

  • Joint calibration of volatility surface and variance term structure
    • Sector: Finance (derivatives), Software
    • What: Implement Yoo-style joint calibration to anchor short-dated options and forward variance simultaneously for cross-maturity stability.
    • Why: Paper confirms robustness of standard methods at short maturities; joint approaches could improve multi-tenor consistency.
    • Tool/product: “Surface+Variance” calibrator with cross-tenor penalties; variance swap integration.
    • Assumptions/dependencies: Access to variance swap or proxy data; careful regularization to avoid overfitting.
  • Enhanced jump models for stressed/event regimes
    • Sector: Finance (equities, FX), Risk
    • What: Upgrade Bates to double-exponential or tempered-stable jumps for better tail control during earnings, macro events, or crises.
    • Why: Paper finds λ≈0 in calm 60-day data; heavy-tail models become valuable in stress.
    • Tool/product: Event-aware pricer that toggles jump families via information criteria and news/event calendars.
    • Assumptions/dependencies: Event detection, dataset breadth across regimes; computational efficiency of new transforms.
  • Multi-factor rate models and cross-asset consistency
    • Sector: Finance (rates, XVA, structured products), Insurance (ALM)
    • What: Extend CIR to multi-factor affine models (e.g., CIR2, Gaussian-affine hybrids) to capture curve shape changes and multi-tenor discounting.
    • Why: Paper’s one-factor CIR fits short-end well; longer horizons and curve shifts need more factors.
    • Tool/product: Multi-factor rate engine with calibration to swaps, OIS, caps/floors; consistent equity-rate joint pricing.
    • Assumptions/dependencies: Rich rate instruments for calibration; careful positivity handling and correlation with equity variance.
  • Unified equity–rate risk engine for XVA and risk transfer
    • Sector: Finance (XVA desks, treasury), Energy/Infrastructure (project finance)
    • What: Integrate Heston/Bates with multi-factor rates to compute CVA/DVA/FVA/MVA under path-dependent discounting and stochastic exposure.
    • Why: Paper quantifies PV differences vs flat DFs; full-stack XVA needs consistent rates and exposure dynamics.
    • Tool/product: Monte Carlo cube with netting sets, collateral, and wrong-way risk flags.
    • Assumptions/dependencies: CSA terms, liquidity metrics, counterparty data, regulatory validation.
  • Real-time, amortized calibration with ML surrogates
    • Sector: Fintech/Software, Market Making
    • What: Train neural surrogates on Fourier-based prices to provide sub-millisecond quote updates while retaining daily anchor calibrations.
    • Why: Paper’s dual-method calibration is robust; ML can emulate pricing fast if trained on accurate labels.
    • Tool/product: “Calibrate-then-Serve” stack: nightly calibration, daytime surrogate inference, drift monitors triggering recalibration.
    • Assumptions/dependencies: Coverage of training domain; guardrails against extrapolation; MRM approval.
  • Dynamic model selection and governance analytics
    • Sector: Finance (model risk), Policy
    • What: A governance layer that statistically tests whether jumps, rough volatility, or stochastic correlation are warranted, switching models as conditions change.
    • Why: Paper suggests jumps unnecessary in current regime; guardrails to adapt when regime shifts.
    • Tool/product: AIC/BIC dashboards, regime detectors (vol-of-vol, skewness/kurtosis metrics), automated approvals.
    • Assumptions/dependencies: Reliable regime indicators; change-control processes; audit trails.
  • Stress testing and supervisory guidance on discounting practices
    • Sector: Policy/Regulation, Banking Supervision
    • What: Issue guidance or stress templates requiring stochastic-rate discounting for longer-dated derivatives to avoid PV bias from flat curves.
    • Why: Paper shows ~2.2% DF difference at 1Y; larger at longer horizons could be material for capital adequacy.
    • Tool/product: Standardized test packs and benchmarks for DF sensitivity to rate dynamics.
    • Assumptions/dependencies: Regulatory consensus; industry implementation timelines.
  • Cross-asset structured products leveraging rate–equity interplay
    • Sector: Structured Products, Wealth
    • What: Design notes whose coupons or barriers adapt to stochastic rate paths (e.g., rate-aware autocallables), priced with unified equity–rate models.
    • Why: Paper’s results imply rates materially impact valuations beyond 1 year; embedding rate dependence can improve risk transfer.
    • Tool/product: Product factory with unified pricing/risk; investor disclosures on rate sensitivity.
    • Assumptions/dependencies: Distribution suitability; secondary market liquidity; robust risk hedging.
  • Curriculum and open-source reference implementation
    • Sector: Academia, Professional Certification
    • What: A complete, open pipeline for Heston/Bates (FFT and Lewis), CIR calibration, MC pricing, and validation notebooks.
    • Why: Paper offers a coherent, reproducible framework useful for training and benchmarking.
    • Tool/product: Open repository with datasets, CI tests, and reproducible examples.
    • Assumptions/dependencies: Data licensing; maintenance/community support.
  • Enterprise-wide ALM integration
    • Sector: Insurance/Banking ALM, Corporate Finance
    • What: Incorporate stochastic rate discounting and short-dated equity volatility modeling into ALM projections, capital planning, and hedge effectiveness testing.
    • Why: Paper quantifies DF impact and supports parsimonious near-term equity dynamics; useful for planning accuracy.
    • Tool/product: ALM simulator with scenario libraries and hedge-optimization routines.
    • Assumptions/dependencies: Data integration across desks; governance of macro-scenarios; computational scaling.

Notes on general feasibility

  • The negligible jump intensity and near-zero vol-of-vol are asset- and period-specific; under stress or for different underlyings, jumps and richer vol dynamics may be essential.
  • CIR’s positivity is desirable but may misfit negative-rate regimes; alternative affine models or shifted dynamics may be required.
  • All results assume risk-neutral pricing with mid-market inputs; bid-ask, liquidity premia, and model risk should be layered in for production.
View Paper Prompt  View All Prompts 

Glossary

  • Affine term-structure model: A class of interest-rate models where bond prices are exponential-affine functions of state variables, enabling tractable term-structure dynamics. "affine term-structure models such as the Cox-Ingersoll-Ross (CIR) process"
  • Arbitrage-free: A property of a model or market precluding riskless profit opportunities; essential for consistent pricing. "essential for arbitrage-free interest-rate modeling"
  • Asian option: An option whose payoff depends on the average price of the underlying over a period. "short-maturity Asian call (~15 days)."
  • At-the-money (ATM): An option with strike equal or close to the current underlying price. "Asian Option Pricing (Monte Carlo, 20-Day ATM)"
  • Basis points (bp): One hundredth of a percent (0.01%), used to measure small changes in rates or errors. "RMSE: 3.65 bp"
  • Bates model: A jump-diffusion extension of Heston that adds discontinuous price jumps to stochastic volatility. "The Bates (1996) model extends the Heston stochastic volatility framework by incorporating discontinuous jumps in the asset price dynamics"
  • Carr-Madan FFT: A fast Fourier transform-based method for option pricing using characteristic functions. "Carr-Madan (1999) FFT"
  • Characteristic function: The Fourier transform of a probability distribution, used to price options in affine models. "Characteristic function of Bates model is"
  • CIR model: The Cox–Ingersoll–Ross square-root process for short-term interest rates with mean reversion and positivity. "The Cox-Ingersoll-Ross (CIR, 1985) model is widely used for risk-neutral evolution of short-term interest rates."
  • Compound Poisson process: A stochastic process with randomly timed jumps whose sizes are random, modeling aggregated jump effects. "reflecting the compound Poisson process of jump arrivals."
  • Cubic spline interpolation: A smooth piecewise-polynomial method to interpolate term structures or curves. "cubic spline interpolation:"
  • Differential Evolution: A population-based global optimization heuristic for continuous parameter calibration. "Differential Evolution (Storn & Price, 1997)"
  • Discount factor: The present-value multiplier applied to a future cash flow under a given interest-rate model. "we compute the discount factors implied by each path:"
  • Double-exponential jumps: A jump-size distribution with double-exponential tails used to better capture skew/kurtosis. "double-exponential jumps, as examined in Agazzotti et al. (2025)"
  • Euribor: Euro Interbank Offered Rate, a benchmark money-market rate used for calibrating interest-rate models. "We calibrate the CIR short-rate model to the Euribor curve (1w-12m)"
  • Feller condition: A parameter restriction in square-root diffusions ensuring strictly positive paths. "This prevents negative rates and is known as the Feller condition."
  • Fourier-based pricing methods: Option valuation approaches that use Fourier transforms of characteristic functions. "Fourier-based pricing methods such as Lewis (2001) inversion and the Carr-Madan (1999) FFT."
  • Fourier inversion: Recovering probabilities or prices by inverting a characteristic function via Fourier integrals. "Lewis (2001) Fourier inversion"
  • Forward-rate scenarios: Simulated future paths of forward interest rates derived from a calibrated model. "generating positive and economically consistent forward-rate scenarios"
  • Implied volatility surface: The collection of implied volatilities across strikes and maturities for an underlying. "the implied volatility surface for this particular stock is nearly flat"
  • Incomplete-market frictions: Real-world market imperfections (e.g., constraints, illiquidity) not captured by complete-market models. "incomplete-market frictions."
  • Jump-diffusion: A process combining continuous diffusion with random jumps to capture sudden price moves. "Jump-diffusion extensions such as the Bates model (Bates, 1996) further account for abrupt price movements"
  • Jump intensity: The average rate at which jumps occur in a jump process. "jump intensities converge to values effectively equal to zero"
  • L-BFGS-B: A limited-memory quasi-Newton algorithm for bound-constrained optimization used in calibration. "L- BFGS-B (Byrd et al., 1995)"
  • Leverage effect: The empirically observed correlation between asset returns and volatility changes. "weak leverage effect (p close to 0)"
  • Log-jump size: The logarithm of the multiplicative jump size, often modeled as normally distributed. "Jt is the log-jump size, assumed Normally distributed"
  • Mean reversion: The tendency of a stochastic process to drift toward a long-run average level. "It enforces positivity of rates, mean reversion, and closed-form expressions for discount factors-properties essential for consistent pricing and risk assessment."
  • Mean-reversion half-life: The time for deviations from the mean to decay by half in a mean-reverting process. "Mean-reversion half-life is t1/2 = In2/k = 0.35 years."
  • Monte Carlo simulation: A numerical method using random sampling to estimate option values or risk metrics. "Monte-Carlo estimates stabilize beyond ~100k paths, consistent with convergence theory (Glasserman, 2004)."
  • Moneyness: A measure of an option’s strike relative to the underlying price (e.g., 95% of spot). "Pricing 70-Day 95% Moneyness Put"
  • Noncentral chi-square distribution: A continuous distribution used for the transition law of the CIR process. "The conditional distribution of IT is noncentral chi-square,"
  • Objective function (MSE): The quantity minimized during calibration (here, mean squared error between model and market prices). "Minimizing the mean squared error (MSE) between model and market prices using the objective function:"
  • Out-of-the-money (OTM): An option with strike unfavorable for immediate exercise (e.g., call with strike above spot). "deep OTM strikes"
  • Poisson process: A stochastic process modeling the random arrival of events (jumps) at a constant average rate. "where Nt is a Poisson process with intensity )"
  • Pricing kernel: A state-price density or stochastic discount factor used in asset pricing under incomplete markets. "alternative pricing kernels"
  • Put-call parity: A relationship linking European call and put prices with the same strike and maturity. "Put prices are incorporated through put-call parity"
  • Risk-neutral measure: A probability measure under which discounted asset prices are martingales for valuation. "Under the risk-neutral measure Q:"
  • Risk-neutral probabilities: Probabilities computed under the risk-neutral measure used in pricing formulas. "where P1 and P2 are risk-neutral probabilities derived from the characteristic function."
  • Rough volatility dynamics: Models where volatility follows a path with rough (non-smooth) behavior, capturing high-frequency features. "rough volatility dynamics"
  • Tempered-stable processes: Lévy processes with tempered heavy tails used for more flexible jump modeling. "tempered-stable processes-may improve the fit"
  • Term structure: The relationship between interest rates or discount factors and maturity. "We calibrate the CIR short-rate model to the Euribor term structure"
  • Variance term structure: The term-structure of variance or volatility used in joint calibration frameworks. "variance-term-structure framework proposed by Yoo (2025)"
  • Volatility-of-volatility: The volatility of the variance process in stochastic volatility models. "volatility-of-volatility parameter collapsed to near zero"
  • Yield curve: The curve plotting interest rates (or yields) across different maturities. "shifts in the yield curve"
  • Zero-coupon bond: A bond paying no coupons and redeemed at face value at maturity; priced via model discounting. "Model zero-coupon bond prices under CIR have closed- form:"
View Paper Prompt  View All Prompts 

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Generate Now

Tweets

Sign up for free to view the 1 tweet with 58 likes about this paper.

Sign Up for Free