Robust deep hedging

Lütkebohmert, Eva; Schmidt, Thorsten; Sester, Julian

doi:10.1080/14697688.2022.2056073

Cited by 12 publications

(6 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the formulation of the robust optimization problem as a worst-case approach, respecting for more distributional ambiguity means to consider more bad scenarios, and therefore may eventually result in a more careful, less volatile trading behavior with smaller returns, as it clearly can be seen in the case ε = 0.3 for the Wasserstein-ball approach and in the case ε = 0.15 for the parametric approach, respectively. In contrast, insufficiently accounting for uncertainty comes at the cost of not being well equipped when adverse scenarios occur, an observation that was already made in similar empirical studies that use different approaches to solve robust optimization problems, compare, e.g., [30] and [32]. Hence, choosing an intermediate level of ambiguity seems to be an appropriate choice.…”

Section: Resultsmentioning

confidence: 99%

Markov Decision Processes under Model Uncertainty

Neufeld¹,

Sester²,

Sikic³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

We introduce a general framework for Markov decision problems under model uncertainty in a discrete-time infinite horizon setting. By providing a dynamic programming principle we obtain a local-to-global paradigm, namely solving a local, i.e., a one time-step robust optimization problem leads to an optimizer of the global (i.e. infinite time-steps) robust stochastic optimal control problem, as well as to a corresponding worst-case measure.Moreover, we apply this framework to portfolio optimization involving data of the S&P 500. We present two different types of ambiguity sets; one is fully data-driven given by a Wasserstein-ball around the empirical measure, the second one is described by a parametric set of multivariate normal distributions, where the corresponding uncertainty sets of the parameters are estimated from the data. It turns out that in scenarios where the market is volatile or bearish, the optimal portfolio strategies from the corresponding robust optimization problem outperforms the ones without model uncertainty, showcasing the importance of taking model uncertainty into account.

show abstract

Section: Resultsmentioning

confidence: 99%

Markov Decision Processes under Model Uncertainty

Neufeld¹,

Sester²,

Sikic³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…– Yield Curve Simulation: The experiments corroborate that the simulated yield curves carry an inductive bias in the Deep ALM framework that affects the learnt strategies. It would therefore be interesting to substitute the HJM-PCA approach with other term structure models and to quantify the model risk of the yield curve simulator; e.g., see Lütkebohmert et al ( 2022 ). Assessing the impact of an exogenously specified non-vanishing market price of risk is equally important.…”

Section: Discussion and Outlookmentioning

confidence: 99%

Deep treasury management for banks

Englisch¹,

Krabichler

Müller³

et al. 2023

Front. Artif. Intell.

View full text Add to dashboard Cite

Retail banks use Asset Liability Management (ALM) to hedge interest rate risk associated with differences in maturity and predictability of their loan and deposit portfolios. The opposing goals of profiting from maturity transformation and hedging interest rate risk while adhering to numerous regulatory constraints make ALM a challenging problem. We formulate ALM as a high-dimensional stochastic control problem in which monthly investment and financing decisions drive the evolution of the bank's balance sheet. To find strategies that maximize long-term utility in the presence of constraints and stochastic interest rates, we train neural networks that parametrize the decision process. Our experiments provide practical insights and demonstrate that the approach of Deep ALM deduces dynamic strategies that outperform static benchmarks.

show abstract

“…In contrast, insufficiently accounting for uncertainty comes at the cost of not being well equipped when adverse scenarios occur, an observation that was already made in similar empirical studies that use different approaches to solve robust optimization problems, compare, for example, Lütkebohmert et al. (2022) and Neufeld et al. (2022).…”

Section: Application To Portfolio Optimizationmentioning

confidence: 99%

“…Wasserstein approach, 𝜆 = 0 𝜀 = 0 0.845952 respecting for more distributional ambiguity means to consider more bad scenarios, and therefore may eventually result in a more careful, less volatile trading behavior with smaller returns, as it clearly can be seen in the case 𝜀 = 0.3 for the Wasserstein-ball approach and in the case 𝜀 = 0.15 for the parametric approach, respectively. In contrast, insufficiently accounting for uncertainty comes at the cost of not being well equipped when adverse scenarios occur, an observation that was already made in similar empirical studies that use different approaches to solve robust optimization problems, compare, for example, Lütkebohmert et al (2022) and Neufeld et al (2022). Hence, choosing an intermediate level of ambiguity seems to be an appropriate choice.…”

Section: Sortino Ratiomentioning

confidence: 99%

Markov decision processes under model uncertainty

Neufeld¹,

Sester

Sikic

2023

Mathematical Finance

Self Cite

View full text Add to dashboard Cite

We introduce a general framework for Markov decision problems under model uncertainty in a discretetime infinite horizon setting. By providing a dynamic programming principle, we obtain a local-to-global paradigm, namely solving a local, that is, a one timestep robust optimization problem leads to an optimizer of the global (i.e., infinite time-steps) robust stochastic optimal control problem, as well as to a corresponding worst-case measure. Moreover, we apply this framework to portfolio optimization involving data of the 𝑆&𝑃 500.We present two different types of ambiguity sets; one is fully data-driven given by a Wasserstein-ball around the empirical measure, the second one is described by a parametric set of multivariate normal distributions, where the corresponding uncertainty sets of the parameters are estimated from the data. It turns out that in scenarios where the market is volatile or bearish, the optimal portfolio strategies from the corresponding robust optimization problem outperforms the ones without model uncertainty, showcasing the importance of taking model uncertainty into account.

show abstract

Robust deep hedging

Cited by 12 publications

References 56 publications

Markov Decision Processes under Model Uncertainty

Markov Decision Processes under Model Uncertainty

Deep treasury management for banks

Markov decision processes under model uncertainty

Contact Info

Product

Resources

About