Search arXiv⌕ Search

arXiv · 2611.09751

Drift-invariant kernels for discount curve models

Abstract

Kernel ridge regression estimates the discount curve as the minimizer of a pricing error plus a penalty given by the norm of a reproducing kernel Hilbert space. By the representer theorem, the estimate lies in a finite-dimensional affine space that depends only on the kernel and on the cash flow dates of the instruments. We ask when this space is invariant under the risk-neutral drift of the discount curve, so that the estimated curves can be the states of a continuous-time arbitrage-free model. We show that this holds for every grid of cash flow dates if and only if the kernel is of the form $k(x,y)=\e^{b(x+y)}\big(c\,\e^{axy}+1-c\big)$, apart from kernels whose Hilbert space has dimension at most two. For this kernel the estimated curve is an affine combination of flat discount curves, each corresponding to a constant continuously compounded yield. The drift-invariant subspaces correspond to sub-grids of the cash flow dates, and the discounted coefficients have linear risk-neutral dynamics.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Andreas Celary, Zehra Eksi-Altay, Damir Filipovic, Paul Krühner. 2026-11-06. Drift-invariant kernels for discount curve models. https://arxiv.org/abs/2611.09751

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

The Long-Only Minimum Variance Portfolio in a One-Factor Market: Theory and Asymptotics

We study the long-only minimum variance (LOMV) portfolio, an object of interest going back to Markowitz, and the simplest non-trivial instance of quantitative portfolio construction. Despite its importance, aspects of the LOMV remain mysterious. Here we examine the base case of a general one-factor covariance model, and provide an explicit characterization of the solution. As a corollary we resolve an open question of Qi (2021) for this setting. In the high-dimensional regime of many assets, we prove that the active ratio (proportion of assets with non-zero weight in the LOMV portfolio) converges in almost all cases to a limit expressed in terms of the distribution of asset betas. In particular, when asset betas are all positive and drawn from a continuous distribution, the active ratio tends to zero in the high dimensional limit, and close to zero when the fraction of negative betas is small.

q-fin.MF↗

On consistency of optimal portfolio choice for state-dependent exponential utilities

In a complete Brownian market, we characterize consistency across investment horizons for state-dependent exponential utilities. We show that consistency is equivalent to an explicit relation between the drift and diffusion of risk tolerance and a deterministic absolutely continuous function. Within this family, the forward performance property selects a unique coefficient. More generally, under the stated admissibility conditions, a forward performance is consistent, whereas consistency alone does not imply the forward performance property.

q-fin.MF↗

Valuation Reveals Uncertainty

This paper studies the recovery of uncertainty from dynamic sublinear valuation rules. Robust valuation assigns each payoff its worst-case expected value over a family of plausible models, thereby inducing a dynamic sublinear valuation rule. While valuation rules are observable in practice, the underlying uncertainty structure remains latent. Under suitable conditions, we show that an uncertainty structure can be recovered from an observed valuation rule and provide an explicit recovery procedure. The construction characterizes the joint uncertainty in the discount factor and the law of the state process.

q-fin.MF↗