Christoph Knochenhauer, Alexander Merkel, Yufei Zhang
We consider the Merton problem of optimizing expected power utility of terminal wealth in the case of an unobservable Markov-modulated drift. What makes the model special is that the agent is allowed to purchase costly expert opinions of varying quality on the current state of the drift, leading to a mixed stochastic control problem with regular and impulse controls involving random consequences. Using ideas from filtering theory, we first embed the original problem with unobservable drift into a full information problem on a larger state space. The value function of the full information problem is characterized as the unique viscosity solution of the dynamic programming PDE. This characterization is achieved by a new variant of the stochastic Perron's method, which additionally allows us to show that, in between purchases of expert opinions, the problem reduces to an exit time control problem which is known to admit an optimal feedback control. Under the assumption of sufficient regularity of this feedback map, we are able to construct optimal trading and expert opinion strategies.
{"title":"Optimal Investment with Costly Expert Opinions","authors":"Christoph Knochenhauer, Alexander Merkel, Yufei Zhang","doi":"arxiv-2409.11569","DOIUrl":"https://doi.org/arxiv-2409.11569","url":null,"abstract":"We consider the Merton problem of optimizing expected power utility of\u0000terminal wealth in the case of an unobservable Markov-modulated drift. What\u0000makes the model special is that the agent is allowed to purchase costly expert\u0000opinions of varying quality on the current state of the drift, leading to a\u0000mixed stochastic control problem with regular and impulse controls involving\u0000random consequences. Using ideas from filtering theory, we first embed the\u0000original problem with unobservable drift into a full information problem on a\u0000larger state space. The value function of the full information problem is\u0000characterized as the unique viscosity solution of the dynamic programming PDE.\u0000This characterization is achieved by a new variant of the stochastic Perron's\u0000method, which additionally allows us to show that, in between purchases of\u0000expert opinions, the problem reduces to an exit time control problem which is\u0000known to admit an optimal feedback control. Under the assumption of sufficient\u0000regularity of this feedback map, we are able to construct optimal trading and\u0000expert opinion strategies.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"49 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142260129","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}
Markowitz laid the foundation of portfolio theory through the mean-variance optimization (MVO) framework. However, the effectiveness of MVO is contingent on the precise estimation of expected returns, variances, and covariances of asset returns, which are typically uncertain. Machine learning models are becoming useful in estimating uncertain parameters, and such models are trained to minimize prediction errors, such as mean squared errors (MSE), which treat prediction errors uniformly across assets. Recent studies have pointed out that this approach would lead to suboptimal decisions and proposed Decision-Focused Learning (DFL) as a solution, integrating prediction and optimization to improve decision-making outcomes. While studies have shown DFL's potential to enhance portfolio performance, the detailed mechanisms of how DFL modifies prediction models for MVO remain unexplored. This study aims to investigate how DFL adjusts stock return prediction models to optimize decisions in MVO, addressing the question: "MSE treats the errors of all assets equally, but how does DFL reduce errors of different assets differently?" Answering this will provide crucial insights into optimal stock return prediction for constructing efficient portfolios.
{"title":"Anatomy of Machines for Markowitz: Decision-Focused Learning for Mean-Variance Portfolio Optimization","authors":"Junhyeong Lee, Inwoo Tae, Yongjae Lee","doi":"arxiv-2409.09684","DOIUrl":"https://doi.org/arxiv-2409.09684","url":null,"abstract":"Markowitz laid the foundation of portfolio theory through the mean-variance\u0000optimization (MVO) framework. However, the effectiveness of MVO is contingent\u0000on the precise estimation of expected returns, variances, and covariances of\u0000asset returns, which are typically uncertain. Machine learning models are\u0000becoming useful in estimating uncertain parameters, and such models are trained\u0000to minimize prediction errors, such as mean squared errors (MSE), which treat\u0000prediction errors uniformly across assets. Recent studies have pointed out that\u0000this approach would lead to suboptimal decisions and proposed Decision-Focused\u0000Learning (DFL) as a solution, integrating prediction and optimization to\u0000improve decision-making outcomes. While studies have shown DFL's potential to\u0000enhance portfolio performance, the detailed mechanisms of how DFL modifies\u0000prediction models for MVO remain unexplored. This study aims to investigate how\u0000DFL adjusts stock return prediction models to optimize decisions in MVO,\u0000addressing the question: \"MSE treats the errors of all assets equally, but how\u0000does DFL reduce errors of different assets differently?\" Answering this will\u0000provide crucial insights into optimal stock return prediction for constructing\u0000efficient portfolios.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"188 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142260130","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}
We use a methodology based on a machine learning algorithm to quantify firms' cyber risks based on their disclosures and a dedicated cyber corpus. The model can identify paragraphs related to determined cyber-threat types and accordingly attribute several related cyber scores to the firm. The cyber scores are unrelated to other firms' characteristics. Stocks with high cyber scores significantly outperform other stocks. The long-short cyber risk factors have positive risk premia, are robust to all factors' benchmarks, and help price returns. Furthermore, we suggest the market does not distinguish between different types of cyber risks but instead views them as a single, aggregate cyber risk.
{"title":"Disentangling the sources of cyber risk premia","authors":"Loïc Maréchal, Nathan Monnet","doi":"arxiv-2409.08728","DOIUrl":"https://doi.org/arxiv-2409.08728","url":null,"abstract":"We use a methodology based on a machine learning algorithm to quantify firms'\u0000cyber risks based on their disclosures and a dedicated cyber corpus. The model\u0000can identify paragraphs related to determined cyber-threat types and\u0000accordingly attribute several related cyber scores to the firm. The cyber\u0000scores are unrelated to other firms' characteristics. Stocks with high cyber\u0000scores significantly outperform other stocks. The long-short cyber risk factors\u0000have positive risk premia, are robust to all factors' benchmarks, and help\u0000price returns. Furthermore, we suggest the market does not distinguish between\u0000different types of cyber risks but instead views them as a single, aggregate\u0000cyber risk.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"215 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142260131","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}
In this research paper, we investigate into a paper named "A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem" [arXiv:1706.10059]. It is a portfolio management problem which is solved by deep learning techniques. The original paper proposes a financial-model-free reinforcement learning framework, which consists of the Ensemble of Identical Independent Evaluators (EIIE) topology, a Portfolio-Vector Memory (PVM), an Online Stochastic Batch Learning (OSBL) scheme, and a fully exploiting and explicit reward function. Three different instants are used to realize this framework, namely a Convolutional Neural Network (CNN), a basic Recurrent Neural Network (RNN), and a Long Short-Term Memory (LSTM). The performance is then examined by comparing to a number of recently reviewed or published portfolio-selection strategies. We have successfully replicated their implementations and evaluations. Besides, we further apply this framework in the stock market, instead of the cryptocurrency market that the original paper uses. The experiment in the cryptocurrency market is consistent with the original paper, which achieve superior returns. But it doesn't perform as well when applied in the stock market.
{"title":"A Deep Reinforcement Learning Framework For Financial Portfolio Management","authors":"Jinyang Li","doi":"arxiv-2409.08426","DOIUrl":"https://doi.org/arxiv-2409.08426","url":null,"abstract":"In this research paper, we investigate into a paper named \"A Deep\u0000Reinforcement Learning Framework for the Financial Portfolio Management\u0000Problem\" [arXiv:1706.10059]. It is a portfolio management problem which is\u0000solved by deep learning techniques. The original paper proposes a\u0000financial-model-free reinforcement learning framework, which consists of the\u0000Ensemble of Identical Independent Evaluators (EIIE) topology, a\u0000Portfolio-Vector Memory (PVM), an Online Stochastic Batch Learning (OSBL)\u0000scheme, and a fully exploiting and explicit reward function. Three different\u0000instants are used to realize this framework, namely a Convolutional Neural\u0000Network (CNN), a basic Recurrent Neural Network (RNN), and a Long Short-Term\u0000Memory (LSTM). The performance is then examined by comparing to a number of\u0000recently reviewed or published portfolio-selection strategies. We have\u0000successfully replicated their implementations and evaluations. Besides, we\u0000further apply this framework in the stock market, instead of the cryptocurrency\u0000market that the original paper uses. The experiment in the cryptocurrency\u0000market is consistent with the original paper, which achieve superior returns.\u0000But it doesn't perform as well when applied in the stock market.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142260132","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}
Frazzini and Pedersen (2014) Betting Against Beta (BAB) factor is based on the idea that high beta assets trade at a premium and low beta assets trade at a discount due to investor funding constraints. However, as argued by Campbell and Vuolteenaho (2004), beta comes in "good" and "bad" varieties. While gaining exposure to low-beta, BAB factors fail to recognize that such a portfolio may tilt towards bad-beta. We propose a Betting Against Bad Beta factor, built by double-sorting on beta and bad-beta and find that it improves the overall performance of BAB strategies though its success relies on proper transaction cost mitigation.
Frazzini 和 Pedersen（2014 年）提出的 "对赌贝塔系数"（BAB）是基于这样一种观点，即由于投资者的资金限制，高贝塔系数的资产会溢价交易，而低贝塔系数的资产会折价交易。然而，正如 Campbell 和 Vuolteenaho（2004 年）所指出的，贝塔系数有 "好 "和 "坏 "之分。在获得低贝塔值投资的同时，BAB 因子未能认识到这种投资组合可能会向坏贝塔值倾斜。我们通过对贝塔系数和坏贝塔系数进行双重排序，提出了一种 "对抗坏贝塔系数"（Betting Against Bad Beta）因子，并发现它能提高 BAB 策略的整体表现，尽管其成功依赖于适当的交易成本缓解。
{"title":"Betting Against (Bad) Beta","authors":"Miguel C. Herculano","doi":"arxiv-2409.00416","DOIUrl":"https://doi.org/arxiv-2409.00416","url":null,"abstract":"Frazzini and Pedersen (2014) Betting Against Beta (BAB) factor is based on\u0000the idea that high beta assets trade at a premium and low beta assets trade at\u0000a discount due to investor funding constraints. However, as argued by Campbell\u0000and Vuolteenaho (2004), beta comes in \"good\" and \"bad\" varieties. While gaining\u0000exposure to low-beta, BAB factors fail to recognize that such a portfolio may\u0000tilt towards bad-beta. We propose a Betting Against Bad Beta factor, built by\u0000double-sorting on beta and bad-beta and find that it improves the overall\u0000performance of BAB strategies though its success relies on proper transaction\u0000cost mitigation.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"55 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142217117","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}
Hamidreza Maleki Almani, Foad Shokrollahi, Tommi Sottinen
We consider the jump-diffusion risky asset model and study its conditional prediction laws. Next, we explain the conditional least square hedging strategy and calculate its closed form for the jump-diffusion model, considering the Black-Scholes framework with interpretations related to investor priorities and transaction costs. We investigate the explicit form of this result for the particular case of the European call option under transaction costs and formulate recursive hedging strategies. Finally, we present a decision tree, table of values, and figures to support our results.
{"title":"Hedging in Jump Diffusion Model with Transaction Costs","authors":"Hamidreza Maleki Almani, Foad Shokrollahi, Tommi Sottinen","doi":"arxiv-2408.10785","DOIUrl":"https://doi.org/arxiv-2408.10785","url":null,"abstract":"We consider the jump-diffusion risky asset model and study its conditional\u0000prediction laws. Next, we explain the conditional least square hedging strategy\u0000and calculate its closed form for the jump-diffusion model, considering the\u0000Black-Scholes framework with interpretations related to investor priorities and\u0000transaction costs. We investigate the explicit form of this result for the\u0000particular case of the European call option under transaction costs and\u0000formulate recursive hedging strategies. Finally, we present a decision tree,\u0000table of values, and figures to support our results.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"421 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142217118","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}
We investigate the optimal investment-reinsurance problem for insurance company with partial information on the market price of the risk. Through the use of filtering techniques we convert the original optimization problem involving different filtrations, into an equivalent stochastic control problem under the observation filtration only, the so-called separated problem. The Markovian structure of the separated problem allows us to apply a classical approach to stochastic optimization based on the Hamilton-Jacobi-Bellman equation, and to provide explicit formulas for the value function and the optimal investment-reinsurance strategy. We finally discuss some comparisons between the optimal strategies pursued by a partially informed insurer and that followed by a fully informed insurer, and we evaluate the value of information using the idea of indifference pricing. These results are also supported by numerical experiments.
{"title":"Portfolio and reinsurance optimization under unknown market price of risk","authors":"Claudia Ceci, Katia Colaneri","doi":"arxiv-2408.07432","DOIUrl":"https://doi.org/arxiv-2408.07432","url":null,"abstract":"We investigate the optimal investment-reinsurance problem for insurance\u0000company with partial information on the market price of the risk. Through the\u0000use of filtering techniques we convert the original optimization problem\u0000involving different filtrations, into an equivalent stochastic control problem\u0000under the observation filtration only, the so-called separated problem. The\u0000Markovian structure of the separated problem allows us to apply a classical\u0000approach to stochastic optimization based on the Hamilton-Jacobi-Bellman\u0000equation, and to provide explicit formulas for the value function and the\u0000optimal investment-reinsurance strategy. We finally discuss some comparisons\u0000between the optimal strategies pursued by a partially informed insurer and that\u0000followed by a fully informed insurer, and we evaluate the value of information\u0000using the idea of indifference pricing. These results are also supported by\u0000numerical experiments.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"34 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142217119","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}
This paper presents a method for accurately predicting the full distribution of stock returns, given a comprehensive set of 194 stock characteristics and market variables. Such distributions, learned from rich data using a machine learning algorithm, are not constrained by restrictive model assumptions and allow the exploration of non-Gaussian, heavy-tailed data and their non-linear interactions. The method uses a two-stage quantile neural network combined with spline interpolation. The results show that the proposed approach outperforms alternative models in terms of out-of-sample losses. Furthermore, we show that the moments derived from such distributions can be useful as alternative empirical estimates in many cases, including mean estimation and forecasting. Finally, we examine the relationship between cross-sectional returns and several distributional characteristics. The results are robust to a wide range of US and international data.
{"title":"Predicting the distributions of stock returns around the globe in the era of big data and learning","authors":"Jozef Barunik, Martin Hronec, Ondrej Tobek","doi":"arxiv-2408.07497","DOIUrl":"https://doi.org/arxiv-2408.07497","url":null,"abstract":"This paper presents a method for accurately predicting the full distribution\u0000of stock returns, given a comprehensive set of 194 stock characteristics and\u0000market variables. Such distributions, learned from rich data using a machine\u0000learning algorithm, are not constrained by restrictive model assumptions and\u0000allow the exploration of non-Gaussian, heavy-tailed data and their non-linear\u0000interactions. The method uses a two-stage quantile neural network combined with\u0000spline interpolation. The results show that the proposed approach outperforms\u0000alternative models in terms of out-of-sample losses. Furthermore, we show that\u0000the moments derived from such distributions can be useful as alternative\u0000empirical estimates in many cases, including mean estimation and forecasting.\u0000Finally, we examine the relationship between cross-sectional returns and\u0000several distributional characteristics. The results are robust to a wide range\u0000of US and international data.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"80 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142217130","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}
This study presents a Reinforcement Learning (RL)-based portfolio management model tailored for high-risk environments, addressing the limitations of traditional RL models and exploiting market opportunities through two-sided transactions and lending. Our approach integrates a new environmental formulation with a Profit and Loss (PnL)-based reward function, enhancing the RL agent's ability in downside risk management and capital optimization. We implemented the model using the Soft Actor-Critic (SAC) agent with a Convolutional Neural Network with Multi-Head Attention (CNN-MHA). This setup effectively manages a diversified 12-crypto asset portfolio in the Binance perpetual futures market, leveraging USDT for both granting and receiving loans and rebalancing every 4 hours, utilizing market data from the preceding 48 hours. Tested over two 16-month periods of varying market volatility, the model significantly outperformed benchmarks, particularly in high-volatility scenarios, achieving higher return-to-risk ratios and demonstrating robust profitability. These results confirm the model's effectiveness in leveraging market dynamics and managing risks in volatile environments like the cryptocurrency market.
{"title":"Optimizing Portfolio with Two-Sided Transactions and Lending: A Reinforcement Learning Framework","authors":"Ali Habibnia, Mahdi Soltanzadeh","doi":"arxiv-2408.05382","DOIUrl":"https://doi.org/arxiv-2408.05382","url":null,"abstract":"This study presents a Reinforcement Learning (RL)-based portfolio management\u0000model tailored for high-risk environments, addressing the limitations of\u0000traditional RL models and exploiting market opportunities through two-sided\u0000transactions and lending. Our approach integrates a new environmental\u0000formulation with a Profit and Loss (PnL)-based reward function, enhancing the\u0000RL agent's ability in downside risk management and capital optimization. We\u0000implemented the model using the Soft Actor-Critic (SAC) agent with a\u0000Convolutional Neural Network with Multi-Head Attention (CNN-MHA). This setup\u0000effectively manages a diversified 12-crypto asset portfolio in the Binance\u0000perpetual futures market, leveraging USDT for both granting and receiving loans\u0000and rebalancing every 4 hours, utilizing market data from the preceding 48\u0000hours. Tested over two 16-month periods of varying market volatility, the model\u0000significantly outperformed benchmarks, particularly in high-volatility\u0000scenarios, achieving higher return-to-risk ratios and demonstrating robust\u0000profitability. These results confirm the model's effectiveness in leveraging\u0000market dynamics and managing risks in volatile environments like the\u0000cryptocurrency market.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"177 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142217131","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}
When constructing portfolios, a key problem is that a lot of financial time series data are sparse, making it challenging to apply machine learning methods. Polymodel theory can solve this issue and demonstrate superiority in portfolio construction from various aspects. To implement the PolyModel theory for constructing a hedge fund portfolio, we begin by identifying an asset pool, utilizing over 10,000 hedge funds for the past 29 years' data. PolyModel theory also involves choosing a wide-ranging set of risk factors, which includes various financial indices, currencies, and commodity prices. This comprehensive selection mirrors the complexities of the real-world environment. Leveraging on the PolyModel theory, we create quantitative measures such as Long-term Alpha, Long-term Ratio, and SVaR. We also use more classical measures like the Sharpe ratio or Morningstar's MRAR. To enhance the performance of the constructed portfolio, we also employ the latest deep learning techniques (iTransformer) to capture the upward trend, while efficiently controlling the downside, using all the features. The iTransformer model is specifically designed to address the challenges in high-dimensional time series forecasting and could largely improve our strategies. More precisely, our strategies achieve better Sharpe ratio and annualized return. The above process enables us to create multiple portfolio strategies aiming for high returns and low risks when compared to various benchmarks.
{"title":"Hedge Fund Portfolio Construction Using PolyModel Theory and iTransformer","authors":"Siqiao Zhao, Zhikang Dong, Zeyu Cao, Raphael Douady","doi":"arxiv-2408.03320","DOIUrl":"https://doi.org/arxiv-2408.03320","url":null,"abstract":"When constructing portfolios, a key problem is that a lot of financial time\u0000series data are sparse, making it challenging to apply machine learning\u0000methods. Polymodel theory can solve this issue and demonstrate superiority in\u0000portfolio construction from various aspects. To implement the PolyModel theory\u0000for constructing a hedge fund portfolio, we begin by identifying an asset pool,\u0000utilizing over 10,000 hedge funds for the past 29 years' data. PolyModel theory\u0000also involves choosing a wide-ranging set of risk factors, which includes\u0000various financial indices, currencies, and commodity prices. This comprehensive\u0000selection mirrors the complexities of the real-world environment. Leveraging on\u0000the PolyModel theory, we create quantitative measures such as Long-term Alpha,\u0000Long-term Ratio, and SVaR. We also use more classical measures like the Sharpe\u0000ratio or Morningstar's MRAR. To enhance the performance of the constructed\u0000portfolio, we also employ the latest deep learning techniques (iTransformer) to\u0000capture the upward trend, while efficiently controlling the downside, using all\u0000the features. The iTransformer model is specifically designed to address the\u0000challenges in high-dimensional time series forecasting and could largely\u0000improve our strategies. More precisely, our strategies achieve better Sharpe\u0000ratio and annualized return. The above process enables us to create multiple\u0000portfolio strategies aiming for high returns and low risks when compared to\u0000various benchmarks.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"52 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141933958","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：