arXiv - QuantFin - Risk Management最新文献

{"title":"Enhancing Deep Hedging of Options with Implied Volatility Surface Feedback Information","authors":"Pascal François, Geneviève Gauthier, Frédéric Godin, Carlos Octavio Pérez Mendoza","doi":"arxiv-2407.21138","DOIUrl":"https://doi.org/arxiv-2407.21138","url":null,"abstract":"We present a dynamic hedging scheme for S&P 500 options, where rebalancing\u0000decisions are enhanced by integrating information about the implied volatility\u0000surface dynamics. The optimal hedging strategy is obtained through a deep\u0000policy gradient-type reinforcement learning algorithm, with a novel hybrid\u0000neural network architecture improving the training performance. The favorable\u0000inclusion of forward-looking information embedded in the volatility surface\u0000allows our procedure to outperform several conventional benchmarks such as\u0000practitioner and smiled-implied delta hedging procedures, both in simulation\u0000and backtesting experiments.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"188 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141862946","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On the optimal design of a new class of proportional portfolio insurance strategies in a jump-diffusion framework","authors":"Katia Colaneri, Daniele Mancinelli, Immacolata Oliva","doi":"arxiv-2407.21148","DOIUrl":"https://doi.org/arxiv-2407.21148","url":null,"abstract":"In this paper, we investigate an optimal investment problem associated with\u0000proportional portfolio insurance (PPI) strategies in the presence of jumps in\u0000the underlying dynamics. PPI strategies enable investors to mitigate downside\u0000risk while still retaining the potential for upside gains. This is achieved by\u0000maintaining an exposure to risky assets proportional to the difference between\u0000the portfolio value and the present value of the guaranteed amount. While PPI\u0000strategies are known to be free of downside risk in diffusion modeling\u0000frameworks with continuous trading, see e.g., Cont and Tankov (2009), real\u0000market applications exhibit a significant non-negligible risk, known as gap\u0000risk, which increases with the multiplier value. The goal of this paper is to\u0000determine the optimal PPI strategy in a setting where gap risk may occur, due\u0000to downward jumps in the asset price dynamics. We consider a loss-averse agent\u0000who aims at maximizing the expected utility of the terminal wealth exceeding a\u0000minimum guarantee. Technically, we model agent's preferences with an S-shaped\u0000utility functions to accommodate the possibility that gap risk occurs, and\u0000address the optimization problem via a generalization of the martingale\u0000approach that turns to be valid under market incompleteness in a jump-diffusion\u0000framework.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"15 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141873427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Design and Optimization of Big Data and Machine Learning-Based Risk Monitoring System in Financial Markets","authors":"Liyang Wang, Yu Cheng, Xingxin Gu, Zhizhong Wu","doi":"arxiv-2407.19352","DOIUrl":"https://doi.org/arxiv-2407.19352","url":null,"abstract":"With the increasing complexity of financial markets and rapid growth in data\u0000volume, traditional risk monitoring methods no longer suffice for modern\u0000financial institutions. This paper designs and optimizes a risk monitoring\u0000system based on big data and machine learning. By constructing a four-layer\u0000architecture, it effectively integrates large-scale financial data and advanced\u0000machine learning algorithms. Key technologies employed in the system include\u0000Long Short-Term Memory (LSTM) networks, Random Forest, Gradient Boosting Trees,\u0000and real-time data processing platform Apache Flink, ensuring the real-time and\u0000accurate nature of risk monitoring. Research findings demonstrate that the\u0000system significantly enhances efficiency and accuracy in risk management,\u0000particularly excelling in identifying and warning against market crash risks.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"31 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141862947","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Set risk measures","authors":"Marcelo Righi, Eduardo Horta, Marlon Moresco","doi":"arxiv-2407.18687","DOIUrl":"https://doi.org/arxiv-2407.18687","url":null,"abstract":"We introduce the concept of set risk measures (SRMs), which are real-valued\u0000maps defined on the space of all non-empty, closed, and bounded sets of almost\u0000surely bounded random variables. Traditional risk measures typically operate on\u0000spaces of random variables, but SRMs extend this framework to sets of random\u0000variables. We establish an axiom scheme for SRMs, similar to classical risk\u0000measures but adapted for set operations. The main technical contribution is an\u0000axiomatic dual representation of convex SRMs by using regular, finitely\u0000additive measures on the unit ball of the dual space of essentially bounded\u0000random variables. We explore worst-case SRMs, which evaluate risk as the\u0000supremum of individual risks within a set, and provide a collection of examples\u0000illustrating the applicability of our framework to systemic risk, portfolio\u0000optimization, and decision-making under uncertainty. This work extends the\u0000theory of risk measures to a more general and flexible setup, accommodating a\u0000broader range of financial and mathematical applications.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"10 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141862948","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Traditional Methods Outperform Generative LLMs at Forecasting Credit Ratings","authors":"Felix Drinkall, Janet B. Pierrehumbert, Stefan Zohren","doi":"arxiv-2407.17624","DOIUrl":"https://doi.org/arxiv-2407.17624","url":null,"abstract":"Large Language Models (LLMs) have been shown to perform well for many\u0000downstream tasks. Transfer learning can enable LLMs to acquire skills that were\u0000not targeted during pre-training. In financial contexts, LLMs can sometimes\u0000beat well-established benchmarks. This paper investigates how well LLMs perform\u0000in the task of forecasting corporate credit ratings. We show that while LLMs\u0000are very good at encoding textual information, traditional methods are still\u0000very competitive when it comes to encoding numeric and multimodal data. For our\u0000task, current LLMs perform worse than a more traditional XGBoost architecture\u0000that combines fundamental and macroeconomic data with high-density text-based\u0000embedding features.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"23 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On the Separability of Vector-Valued Risk Measures","authors":"Çağın Ararat, Zachary Feinstein","doi":"arxiv-2407.16878","DOIUrl":"https://doi.org/arxiv-2407.16878","url":null,"abstract":"Risk measures for random vectors have been considered in multi-asset markets\u0000with transaction costs and financial networks in the literature. While the\u0000theory of set-valued risk measures provide an axiomatic framework for assigning\u0000to a random vector its set of all capital requirements or allocation vectors,\u0000the actual decision-making process requires an additional rule to select from\u0000this set. In this paper, we define vector-valued risk measures by an analogous\u0000list of axioms and show that, in the convex and lower semicontinuous case, such\u0000functionals always ignore the dependence structures of the input random\u0000vectors. We also show that set-valued risk measures do not have this issue as\u0000long as they do not reduce to a vector-valued functional. Finally, we\u0000demonstrate that our results also generalize to the conditional setting. These\u0000results imply that convex vector-valued risk measures are not suitable for\u0000defining capital allocation rules for a wide range of financial applications\u0000including systemic risk measures.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"14 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"On Deep Learning for computing the Dynamic Initial Margin and Margin Value Adjustment","authors":"Joel P. Villarino, Álvaro Leitao","doi":"arxiv-2407.16435","DOIUrl":"https://doi.org/arxiv-2407.16435","url":null,"abstract":"The present work addresses the challenge of training neural networks for\u0000Dynamic Initial Margin (DIM) computation in counterparty credit risk, a task\u0000traditionally burdened by the high costs associated with generating training\u0000datasets through nested Monte Carlo (MC) simulations. By condensing the initial\u0000market state variables into an input vector, determined through an interest\u0000rate model and a parsimonious parameterization of the current interest rate\u0000term structure, we construct a training dataset where labels are noisy but\u0000unbiased DIM samples derived from single MC paths. A multi-output neural\u0000network structure is employed to handle DIM as a time-dependent function,\u0000facilitating training across a mesh of monitoring times. The methodology offers\u0000significant advantages: it reduces the dataset generation cost to a single MC\u0000execution and parameterizes the neural network by initial market state\u0000variables, obviating the need for repeated training. Experimental results\u0000demonstrate the approach's convergence properties and robustness across\u0000different interest rate models (Vasicek and Hull-White) and portfolio\u0000complexities, validating its general applicability and efficiency in more\u0000realistic scenarios.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"21 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778291","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Analyzing selected cryptocurrencies spillover effects on global financial indices: Comparing risk measures using conventional and eGARCH-EVT-Copula approaches","authors":"Shafique Ur Rehman, Touqeer Ahmad, Wu Dash Desheng, Amirhossein Karamoozian","doi":"arxiv-2407.15766","DOIUrl":"https://doi.org/arxiv-2407.15766","url":null,"abstract":"This study examines the interdependence between cryptocurrencies and\u0000international financial indices, such as MSCI World and MSCI Emerging Markets.\u0000We compute the value at risk, expected shortfall (ES), and range value at risk\u0000(RVaR) and investigate the dynamics of risk spillover. We employ a hybrid\u0000approach to derive these risk measures that integrate GARCH models, extreme\u0000value models, and copula functions. This framework uses a bivariate portfolio\u0000approach involving cryptocurrency data and traditional financial indices. To\u0000estimate the above risks of these portfolio structures, we employ symmetric and\u0000asymmetric GARCH and both tail flexible EVT models as marginal to model the\u0000marginal distribution of each return series and apply different copula\u0000functions to connect the pairs of marginal distributions into a multivariate\u0000distribution. The empirical findings indicate that the eGARCH EVT-based copula\u0000model adeptly captures intricate dependencies, surpassing conventional\u0000methodologies like Historical simulations and t-distributed parametric in VaR\u0000estimation. At the same time, the HS method proves superior for ES, and the\u0000t-distributed parametric method outperforms RVaR. Eventually, the\u0000Diebold-Yilmaz approach will be applied to compute risk spillovers between four\u0000sets of asset sequences. This phenomenon implies that cryptocurrencies reveal\u0000substantial spillover effects among themselves but minimal impact on other\u0000assets. From this, it can be concluded that cryptocurrencies propose\u0000diversification benefits and do not provide hedging advantages within an\u0000investor's portfolio. Our results underline RVaR superiority over ES regarding\u0000regulatory arbitrage and model misspecification. The conclusions of this study\u0000will benefit investors and financial market professionals who aspire to\u0000comprehend digital currencies as a novel asset class and attain perspicuity in\u0000regulatory arbitrage.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"62 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Counter-monotonic risk allocations and distortion risk measures","authors":"Mario Ghossoub, Qinghua Ren, Ruodu Wang","doi":"arxiv-2407.16099","DOIUrl":"https://doi.org/arxiv-2407.16099","url":null,"abstract":"In risk-sharing markets with aggregate uncertainty, characterizing\u0000Pareto-optimal allocations when agents might not be risk averse is a\u0000challenging task, and the literature has only provided limited explicit results\u0000thus far. In particular, Pareto optima in such a setting may not necessarily be\u0000comonotonic, in contrast to the case of risk-averse agents. In fact, when\u0000market participants are risk-seeking, Pareto-optimal allocations are\u0000counter-monotonic. Counter-monotonicity of Pareto optima also arises in some\u0000situations for quantile-optimizing agents. In this paper, we provide a\u0000systematic study of efficient risk sharing in markets where allocations are\u0000constrained to be counter-monotonic. The preferences of the agents are modelled\u0000by a common distortion risk measure, or equivalently, by a common Yaari dual\u0000utility. We consider three different settings: risk-averse agents, risk-seeking\u0000agents, and those with an inverse S-shaped distortion function. In each case,\u0000we provide useful characterizations of optimal allocations, for both the\u0000counter-monotonic market and the unconstrained market. To illustrate our\u0000results, we consider an application to a portfolio choice problem for a\u0000portfolio manager tasked with managing the investments of a group of clients,\u0000with varying levels of risk aversion or risk seeking. We determine explicitly\u0000the optimal investment strategies in this case. Our results confirm the\u0000intuition that a manager investing on behalf of risk-seeking agents tends to\u0000invest more in risky assets than a manager acting on behalf of risk-averse\u0000agents.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"16 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778289","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Is the difference between deep hedging and delta hedging a statistical arbitrage?","authors":"Pascal François, Geneviève Gauthier, Frédéric Godin, Carlos Octavio Pérez Mendoza","doi":"arxiv-2407.14736","DOIUrl":"https://doi.org/arxiv-2407.14736","url":null,"abstract":"The recent work of Horikawa and Nakagawa (2024) explains that there exist\u0000complete market models in which the difference between the hedging position\u0000provided by deep hedging and that of the replicating portfolio is a statistical\u0000arbitrage. This raises concerns as it entails that deep hedging can include a\u0000speculative component aimed simply at exploiting the structure of the risk\u0000measure guiding the hedging optimisation problem. We test whether such finding\u0000remains true in a GARCH-based market model. We observe that the difference\u0000between deep hedging and delta hedging can be a statistical arbitrage if the\u0000risk measure considered does not put sufficient relative weight on adverse\u0000outcomes. Nevertheless, a suitable choice of risk measure can prevent the deep\u0000hedging agent from including a speculative overlay within its hedging strategy.","PeriodicalId":501128,"journal":{"name":"arXiv - QuantFin - Risk Management","volume":"411 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141778324","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}