Daily Summary for 2026-11-11

daily

Transcript

Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.

Ruth: It's the eleventh of November, twenty twenty-six, and this is the day's research.

Warren: 4 new papers came out today.

Ruth: I'm Ruth.

Warren: And I'm Warren, here with Iris.

Ruth: We'll take the day in one pass, then pull out the papers we're staying with.

Warren: Start us off with Implied Volatility Expansions for VIX Options in Forward Variance Models.

The day's research: Ruth: We are diving into discovering SDE dynamics from one path without ergodicity assumptions.

Warren: The work matters is building a data-driven framework for model discovery from a single trajectory.

Iris: That framework bypasses restrictive assumptions like stationarity in standard models.

Ruth: It combines stochastic Taylor expansions with Girsanov transformations to estimate drift and noise simultaneously.

Warren: Building on those estimators, the first sparse identification algorithm identifies dynamics from a single observation.

Iris: This avoids the need for ergodicity or stationarity when identifying governing SDEs from one observation.

Ruth: That leads into formalizing mathematical finance with high reliability using a verified library in Lean 4.

Warren: That library rigorously checks financial mathematics, including constructing the Itô integral within the environment.

Iris: The work also addresses pricing and hedging in complex settings using Hilbert space structures for infinite-dimensional dynamics.

Ruth: That framework extends Black-Scholes concepts to markets with infinite-dimensional asset price dynamics.

Warren: We have research into implied volatility expansions for VIX options in forward variance models today.

Iris: New techniques provide explicit formulas achieving single-kernel model accuracy in those expansions.

Lucky paper: 2604.25123: Ruth: Okay, let's look at this paper now, Implied Volatility Expansions for VIX Options in Forward Variance Models.

Warren: So the main goal here is to get closed-form expansions for implied volatility without needing numerical root-finding using option prices.

Iris: It focuses on VIX options within forward variance models, which are built around the term structure of implied variances of log-contracts.

Ruth: They define an instantaneous forward variance curve, ξu t equals d/du (u minus t) two sigma t u, and then solve that to get a log-normal process for the instantaneous log-forward variance Xu t.

Warren: So they derive an explicit implied volatility expansion from the weak VIX option price approximation of ten.

Iris: That approximation involves an expectation term E phi with a remainder term bounded by C times the cube of two alpha one and alpha two.

Ruth: And they approximate the VIX implied volatility, σBS(k, T), using terms like sigma P squared plus gamma two squared sigma P T plus three gamma three over eight sigma P T minus gamma three over sigma cubed P T squared times (xP minus k).

Iris: They then extend this construction to mixed-kernel forward variance models where the forward variance process is a convex mixture of two log-normal components.

Ruth: In that mixed case, they get an approximation for the VIX implied volatility that involves terms like sigma P sub j squared plus p lambda j e xP sub j e x theta sub j root T times X N n times ci over i nu P one omega n Hen minus one alpha minus sigma P sub j root T squared.

Warren: So numerically, they show the expansions perform well with relative errors below one percent for VIX maturities T equal to one three and six months.

Iris: They also note that the implied volatility expansion is about one hundred thirty-one times faster than the quadrature method in the standard Bergomi model.

Ruth: And for calibration tests, they found that the implied volatility expansion in equation number forty captures market smiles quite well when calibrated to market data for the mixed rough Bergomi model.

Warren: They also say that calibration results obtained from their implied volatility expansion can be regarded as reliable when compared to those from the weak price approximation.

Iris: The limitation they mention is in the mixed-kernel case, where the Hermite-based expansion captures market VIX skews more effectively only if the two proxy volatilities are not too far apart.

Ruth: So if sigma P sub one and sigma P sub two are far from each other, accuracy can drop in that mixed-kernel setting.

Warren: In more extreme parameter regimes, they admit the expansion might lose accuracy but still provides reliable initial guesses for standard root-finding procedures.

Iris: They also found that in the mixed standard Bergomi model, the performance of equation number forty-five is slightly worse than equation number forty when omega one and omega two are close.

Ruth: And they noted that their runtimes for the implied volatility expansion are comparable to those of the weak approximation.

Warren: They also have additional calibration tests showing that the calibrated smiles reported in figure number ten show that the implied volatility expansion formula number forty captures market smiles quite well.

More episodes

← Home