arXiv - QuantFin - Portfolio Management最新文献

{"title":"Recommender Systems in Financial Trading: Using machine-based conviction analysis in an explainable AI investment framework","authors":"Alicia Vidler","doi":"arxiv-2404.11080","DOIUrl":"https://doi.org/arxiv-2404.11080","url":null,"abstract":"Traditionally, assets are selected for inclusion in a portfolio (long or\u0000short) by human analysts. Teams of human portfolio managers (PMs) seek to weigh\u0000and balance these securities using optimisation methods and other portfolio\u0000construction processes. Often, human PMs consider human analyst recommendations\u0000against the backdrop of the analyst's recommendation track record and the\u0000applicability of the analyst to the recommendation they provide. Many firms\u0000regularly ask analysts to provide a \"conviction\" level on their\u0000recommendations. In the eyes of PMs, understanding a human analyst's track\u0000record has typically come down to basic spread sheet tabulation or, at best, a\u0000\"virtual portfolio\" paper trading book to keep track of results of\u0000recommendations. Analysts' conviction around their recommendations and their \"paper trading\"\u0000track record are two crucial workflow components between analysts and portfolio\u0000construction. Many human PMs may not even appreciate that they factor these\u0000data points into their decision-making logic. This chapter explores how\u0000Artificial Intelligence (AI) can be used to replicate these two steps and\u0000bridge the gap between AI data analytics and AI-based portfolio construction\u0000methods. This field of AI is referred to as Recommender Systems (RS). This\u0000chapter will further explore what metadata that RS systems functionally supply\u0000to downstream systems and their features.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"39 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140617698","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":"Developing An Attention-Based Ensemble Learning Framework for Financial Portfolio Optimisation","authors":"Zhenglong Li, Vincent Tam","doi":"arxiv-2404.08935","DOIUrl":"https://doi.org/arxiv-2404.08935","url":null,"abstract":"In recent years, deep or reinforcement learning approaches have been applied\u0000to optimise investment portfolios through learning the spatial and temporal\u0000information under the dynamic financial market. Yet in most cases, the existing\u0000approaches may produce biased trading signals based on the conventional price\u0000data due to a lot of market noises, which possibly fails to balance the\u0000investment returns and risks. Accordingly, a multi-agent and self-adaptive\u0000portfolio optimisation framework integrated with attention mechanisms and time\u0000series, namely the MASAAT, is proposed in this work in which multiple trading\u0000agents are created to observe and analyse the price series and directional\u0000change data that recognises the significant changes of asset prices at\u0000different levels of granularity for enhancing the signal-to-noise ratio of\u0000price series. Afterwards, by reconstructing the tokens of financial data in a\u0000sequence, the attention-based cross-sectional analysis module and temporal\u0000analysis module of each agent can effectively capture the correlations between\u0000assets and the dependencies between time points. Besides, a portfolio generator\u0000is integrated into the proposed framework to fuse the spatial-temporal\u0000information and then summarise the portfolios suggested by all trading agents\u0000to produce a newly ensemble portfolio for reducing biased trading actions and\u0000balancing the overall returns and risks. The experimental results clearly\u0000demonstrate that the MASAAT framework achieves impressive enhancement when\u0000compared with many well-known portfolio optimsation approaches on three\u0000challenging data sets of DJIA, S&P 500 and CSI 300. More importantly, our\u0000proposal has potential strengths in many possible applications for future\u0000study.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"48 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140601645","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":"Quantum computing approach to realistic ESG-friendly stock portfolios","authors":"Francesco Catalano, Laura Nasello, Daniel Guterding","doi":"arxiv-2404.02582","DOIUrl":"https://doi.org/arxiv-2404.02582","url":null,"abstract":"Finding an optimal balance between risk and returns in investment portfolios\u0000is a central challenge in quantitative finance, often addressed through\u0000Markowitz portfolio theory (MPT). While traditional portfolio optimization is\u0000carried out in a continuous fashion, as if stocks could be bought in fractional\u0000increments, practical implementations often resort to approximations, as\u0000fractional stocks are typically not tradeable. While these approximations are\u0000effective for large investment budgets, they deteriorate as budgets decrease.\u0000To alleviate this issue, a discrete Markowitz portfolio theory (DMPT) with\u0000finite budgets and integer stock weights can be formulated, but results in a\u0000non-polynomial (NP)-hard problem. Recent progress in quantum processing units\u0000(QPUs), including quantum annealers, makes solving DMPT problems feasible. Our\u0000study explores portfolio optimization on quantum annealers, establishing a\u0000mapping between continuous and discrete Markowitz portfolio theories. We find\u0000that correctly normalized discrete portfolios converge to continuous solutions\u0000as budgets increase. Our DMPT implementation provides efficient frontier\u0000solutions, outperforming traditional rounding methods, even for moderate\u0000budgets. Responding to the demand for environmentally and socially responsible\u0000investments, we enhance our discrete portfolio optimization with ESG\u0000(environmental, social, governance) ratings for EURO STOXX 50 index stocks. We\u0000introduce a utility function incorporating ESG ratings to balance risk, return,\u0000and ESG-friendliness, and discuss implications for ESG-aware investors.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"95 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140601753","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":"Using Machine Learning to Forecast Market Direction with Efficient Frontier Coefficients","authors":"Nolan Alexander, William Scherer","doi":"arxiv-2404.00825","DOIUrl":"https://doi.org/arxiv-2404.00825","url":null,"abstract":"We propose a novel method to improve estimation of asset returns for\u0000portfolio optimization. This approach first performs a monthly directional\u0000market forecast using an online decision tree. The decision tree is trained on\u0000a novel set of features engineered from portfolio theory: the efficient\u0000frontier functional coefficients. Efficient frontiers can be decomposed to\u0000their functional form, a square-root second-order polynomial, and the\u0000coefficients of this function captures the information of all the constituents\u0000that compose the market in the current time period. To make these forecasts\u0000actionable, these directional forecasts are integrated to a portfolio\u0000optimization framework using expected returns conditional on the market\u0000forecast as an estimate for the return vector. This conditional expectation is\u0000calculated using the inverse Mills ratio, and the Capital Asset Pricing Model\u0000is used to translate the market forecast to individual asset forecasts. This\u0000novel method outperforms baseline portfolios, as well as other feature sets\u0000including technical indicators and the Fama-French factors. To empirically\u0000validate the proposed model, we employ a set of market sector ETFs.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"36 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140601666","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":"Portfolio management using graph centralities: Review and comparison","authors":"Bahar Arslan, Vanni Noferini, Spyridon Vrontos","doi":"arxiv-2404.00187","DOIUrl":"https://doi.org/arxiv-2404.00187","url":null,"abstract":"We investigate an application of network centrality measures to portfolio\u0000optimization, by generalizing the method in [Pozzi, Di Matteo and Aste,\u0000emph{Spread of risks across financial markets: better to invest in the\u0000peripheries}, Scientific Reports 3:1665, 2013], that however had significant\u0000limitations with respect to the state of the art in network theory. In this\u0000paper, we systematically compare many possible variants of the originally\u0000proposed method on S&P 500 stocks. We use daily data from twenty-seven years\u0000as training set and their following year as test set. We thus select the best\u0000network-based methods according to different viewpoints including for instance\u0000the highest Sharpe Ratio and the highest expected return. We give emphasis in\u0000new centrality measures and we also conduct a thorough analysis, which reveals\u0000significantly stronger results compared to those with more traditional methods.\u0000According to our analysis, this graph-theoretical approach to investment can be\u0000used successfully by investors with different investment profiles leading to\u0000high risk-adjusted returns.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"95 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140601637","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":"Rank-Dependent Predictable Forward Performance Processes","authors":"Bahman Angoshtari, Shida Duan","doi":"arxiv-2403.16228","DOIUrl":"https://doi.org/arxiv-2403.16228","url":null,"abstract":"Predictable forward performance processes (PFPPs) are stochastic optimal\u0000control frameworks for an agent who controls a randomly evolving system but can\u0000only prescribe the system dynamics for a short period ahead. This is a common\u0000scenario in which a controlling agent frequently re-calibrates her model. We\u0000introduce a new class of PFPPs based on rank-dependent utility, generalizing\u0000existing models that are based on expected utility theory (EUT). We establish\u0000existence of rank-dependent PFPPs under a conditionally complete market and\u0000exogenous probability distortion functions which are updated periodically. We\u0000show that their construction reduces to solving an integral equation that\u0000generalizes the integral equation obtained under EUT in previous studies. We\u0000then propose a new approach for solving the integral equation via theory of\u0000Volterra equations. We illustrate our result in the special case of\u0000conditionally complete Black-Scholes model.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"57 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140300947","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":"DiffSTOCK: Probabilistic relational Stock Market Predictions using Diffusion Models","authors":"Divyanshu Daiya, Monika Yadav, Harshit Singh Rao","doi":"arxiv-2403.14063","DOIUrl":"https://doi.org/arxiv-2403.14063","url":null,"abstract":"In this work, we propose an approach to generalize denoising diffusion\u0000probabilistic models for stock market predictions and portfolio management.\u0000Present works have demonstrated the efficacy of modeling interstock relations\u0000for market time-series forecasting and utilized Graph-based learning models for\u0000value prediction and portfolio management. Though convincing, these\u0000deterministic approaches still fall short of handling uncertainties i.e., due\u0000to the low signal-to-noise ratio of the financial data, it is quite challenging\u0000to learn effective deterministic models. Since the probabilistic methods have\u0000shown to effectively emulate higher uncertainties for time-series predictions.\u0000To this end, we showcase effective utilisation of Denoising Diffusion\u0000Probabilistic Models (DDPM), to develop an architecture for providing better\u0000market predictions conditioned on the historical financial indicators and\u0000inter-stock relations. Additionally, we also provide a novel deterministic\u0000architecture MaTCHS which uses Masked Relational Transformer(MRT) to exploit\u0000inter-stock relations along with historical stock features. We demonstrate that\u0000our model achieves SOTA performance for movement predication and Portfolio\u0000management.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"30 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140198013","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":"Asset management with an ESG mandate","authors":"Michele Azzone, Emilio Barucci, Davide Stocco","doi":"arxiv-2403.11622","DOIUrl":"https://doi.org/arxiv-2403.11622","url":null,"abstract":"We investigate the portfolio frontier and risk premia in equilibrium when an\u0000institutional investor aims to minimize the tracking error variance and to\u0000attain an ESG score higher than the benchmark's one (ESG mandate). Provided\u0000that a negative ESG premium for stocks is priced by the market, we show that an\u0000ESG mandate can reduce the mean-variance inefficiency of the portfolio frontier\u0000when the asset manager targets a limited over-performance with respect to the\u0000benchmark. Instead, for a high over-performance target, an ESG mandate leads to\u0000a higher variance. The mean-variance improvement is due to the fact that the\u0000ESG mandate induces the asset manager to over-invest in assets with a high\u0000mean-standard deviation ratio. In equilibrium, with asset managers and\u0000mean-variance investors, a negative ESG premium arises if the ESG mandate is\u0000binding for asset managers. A result that is borne out by the data.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"163 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140170861","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":"Can a GPT4-Powered AI Agent Be a Good Enough Performance Attribution Analyst?","authors":"Bruno de Melo","doi":"arxiv-2403.10482","DOIUrl":"https://doi.org/arxiv-2403.10482","url":null,"abstract":"Performance attribution analysis, defined as the process of explaining the\u0000drivers of the excess performance of an investment portfolio against a\u0000benchmark, stands as a significant aspect of portfolio management and plays a\u0000crucial role in the investment decision-making process, particularly within the\u0000fund management industry. Rooted in a solid financial and mathematical\u0000framework, the importance and methodologies of this analytical technique are\u0000extensively documented across numerous academic research papers and books. The\u0000integration of large language models (LLMs) and AI agents marks a\u0000groundbreaking development in this field. These agents are designed to automate\u0000and enhance the performance attribution analysis by accurately calculating and\u0000analyzing portfolio performances against benchmarks. In this study, we\u0000introduce the application of an AI Agent for a variety of essential performance\u0000attribution tasks, including the analysis of performance drivers and utilizing\u0000LLMs as calculation engine for multi-level attribution analysis and\u0000question-answer (QA) exercises. Leveraging advanced prompt engineering\u0000techniques such as Chain-of-Thought (CoT) and Plan and Solve (PS), and\u0000employing a standard agent framework from LangChain, the research achieves\u0000promising results: it achieves accuracy rates exceeding 93% in analyzing\u0000performance drivers, attains 100% in multi-level attribution calculations, and\u0000surpasses 84% accuracy in QA exercises that simulate official examination\u0000standards. These findings affirm the impactful role of AI agents, prompt\u0000engineering and evaluation in advancing portfolio management processes,\u0000highlighting a significant advancement in the practical application and\u0000evaluation of AI technologies within the domain.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"106 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140146356","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":"An Analytical Approach to (Meta)Relational Models Theory, and its Application to Triple Bottom Line (Profit, People, Planet) -- Towards Social Relations Portfolio Management","authors":"Arsham Farzinnia, Corine Boon","doi":"arxiv-2402.18764","DOIUrl":"https://doi.org/arxiv-2402.18764","url":null,"abstract":"Investigating the optimal nature of social interactions among generic actors\u0000(e.g., people or firms), aiming to achieve specifically-agreed objectives, has\u0000been the subject of extensive academic research. Using the relational models\u0000theory - comprehensively describing all social interactions among actors as\u0000combinations of only four forms of sociality: communal sharing, authority\u0000ranking, equality matching, and market pricing - the common approach within the\u0000literature revolves around qualitative assessments of the sociality models'\u0000configurations most effective in realizing predefined purposes, at times\u0000supplemented by empirical data. In this treatment, we formulate this question\u0000as a mathematical optimization problem, in order to quantitatively determine\u0000the best possible configurations of sociality forms between dyadic actors which\u0000would optimize their mutually-agreed objectives. For this purpose, we develop\u0000an analytical framework for quantifying the (meta)relational models theory, and\u0000mathematically demonstrate that combining the four sociality forms within a\u0000specific meaningful social interaction inevitably prompts an inherent tension\u0000among them, through a single elementary and universal metarelation. In analogy\u0000with financial portfolio management, we subsequently introduce the concept of\u0000Social Relations Portfolio (SRP) management, and propose a generalizable\u0000procedural methodology capable of quantitatively identifying the efficient SRP\u0000for any objective involving meaningful social relations. As an important\u0000illustration, the methodology is applied to the Triple Bottom Line paradigm to\u0000derive its efficient SRP, guiding practitioners in precisely measuring,\u0000monitoring, reporting and (proactively) steering stakeholder management efforts\u0000regarding Corporate Social Responsibility (CSR) and Environmental, Social and\u0000Governance (ESG) within and / or across organizations.","PeriodicalId":501045,"journal":{"name":"arXiv - QuantFin - Portfolio Management","volume":"36 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-02-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140006495","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}