Andrey Itkin

Igor Halperin, Andrey Itkin

This paper introduces a dynamic portfolio optimization framework for large institutional investors using Scientific Physics-Informed Reinforcement Learning (SciPhyRL). Formulated in continuous time over an extended state space that includes explicit cumulative costs, the approach leverages offline historical data to learn optimal, distribution-aware strategies. A core innovation reduces the optimization challenge to solving an HJB equation by projecting it onto observed trajectories as a pathwise Hamilton-Jacobi equation. This is solved directly from data using PINN in a single offline sweep, eliminating the need for traditional value or policy iteration. To make the method effective at practical short horizons, the control variable is recast from a continuous trading rate to a discrete target holding. This ensures signal-implied positions are reached immediately, while execution costs are evaluated against a microstructure-grounded quadratic price impact model. Evaluated on a $14$-asset ETF universe using an engineered oracle signal, the learned Gibbs policy yields substantial out-of-sample Sharpe ratio improvements over static and myopic baselines. The results demonstrate that the proposed framework successfully translates known signal quality into a robust, multi-period, and cost-aware allocation mechanism with strictly controlled volatility and turnover.

Andrey Itkin

The Fokker-Planck equation is fundamental to statistical mechanics, yet in settings with multiple state variables, anisotropic (cross-) diffusion, and jumps, conventional discretizations frequently produce non-physical negative probability densities. Building on the operator approach of "A. Itkin, Pricing derivatives under Levy models. Modern finite difference and pseudo-differential operators approach, Springer, 2017, ISBN 978-1-4939-6792-6", we introduce a family of "Diagonal Frog" discretizations whose spatial operators are eventually M-matrices (EM-matrices). Although these operators lack a local M-matrix structure, positivity of the directional sub-operators emerges in the spirit of Zeno's paradox: the matrix exponential, assembled as the limit of infinitely many ever-smaller substeps, is provably nonnegative after a short transient even though no single substep is. For the mixed-derivative block, whose generator is not eventually nonnegative, positivity instead rests on a factorized resolvent solver and holds conditionally, on an explicit step-size window; discrete mass is conserved exactly by the splitting for every step size. The resulting schemes are second-order accurate in time and space and require O(m 2 N + m 3) operations per time step, where m is the dimension of the Krylov subspace used to apply the exponential. As stress tests, we solve a two-dimensional anisotropic Fokker-Planck equation in the strong cross-diffusion regime against an exact Gaussian reference, a Kramers escape problem in a double-well potential, and an advection-dominated problem, and observe that the schemes remain stable, nonnegative, and mass-conservative for a wide range of Pécklet numbers (so, don't need any flux limiter). Finally, we extend the construction to multidimensional processes and to the backward Kolmogorov equation with jumps.

Leif Andersen, Andrey Itkin, Rakhymzhan Kazbek

A flexible forward (FF) is a customized FX hedging instrument that guarantees a fixed exchange rate while letting the holder choose the delivery date within a pre-agreed window. It is therefore an American-style option on timing, and its valuation must respect the volatility skew of the underlying currency pair. We price FF contracts (and, more generally, American options) under a time-inhomogeneous Heston model which captures the forward-skew term structure while preserving analytical tractability through a recursive (matrix) Riccati solution for the joint characteristic function. Extending the integral-equation (decomposition) approach to time-dependent coefficients, we derive a Volterra equation characterizing the early-exercise surface. The expectation in the decomposition formula is evaluated by two complementary spectral methods: a double cosine (COS) expansion of the transition density, and a damped-Sinc (DSINC) local-basis scheme that is more accurate and stays robust when a low Feller ratio or large vol-of-vol induces Gibbs oscillations in the COS series. Benchmarked against a penalty-iteration MCS-ADI finite-difference solver, both methods price a contract in about 1-2 seconds, roughly an order of magnitude faster than the finest finite-difference grid, while DSINC improves median accuracy over COS by about a factor of twelve. The experiments also show that the early-exercise surface is a substantially nonlinear function of the variance, contrary to the linear-in-variance approximation common in earlier work.