|
on Big Data |
| By: | Adamantios Ntakaris; Gbenga Ibikunle |
| Abstract: | High-frequency trading (HFT) has transformed modern financial markets, making reliable short-term price forecasting models essential. In this study, we present a novel approach to mid-price forecasting using Level 1 limit order book (LOB) data from NASDAQ, focusing on 100 U.S. stocks from the S&P 500 index during the period from September to November 2022. Expanding on our previous work with Radial Basis Function Neural Networks (RBFNN), which leveraged automated feature importance techniques based on mean decrease impurity (MDI) and gradient descent (GD), we introduce the Adaptive Learning Policy Engine (ALPE) - a reinforcement learning (RL)-based agent designed for batch-free, immediate mid-price forecasting. ALPE incorporates adaptive epsilon decay to dynamically balance exploration and exploitation, outperforming a diverse range of highly effective machine learning (ML) and deep learning (DL) models in forecasting performance. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.19372 |
| By: | Kamil {\L}. Szyd{\l}owski; Jaros{\l}aw A. Chudziak |
| Abstract: | This paper investigates the application of Transformer-based neural networks to stock price forecasting, with a special focus on the intersection of machine learning techniques and financial market analysis. The evolution of Transformer models, from their inception to their adaptation for time series analysis in financial contexts, is reviewed and discussed. Central to our study is the exploration of the Hidformer model, which is currently recognized for its promising performance in time series prediction. The primary aim of this paper is to determine whether Hidformer will also prove itself in the task of stock price prediction. This slightly modified model serves as the framework for our experiments, integrating the principles of technical analysis with advanced machine learning concepts to enhance stock price prediction accuracy. We conduct an evaluation of the Hidformer model's performance, using a set of criteria to determine its efficacy. Our findings offer additional insights into the practical application of Transformer architectures in financial time series forecasting, highlighting their potential to improve algorithmic trading strategies, including human decision making. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.19932 |
| By: | Sermet Pekin; Aykut Sengul |
| Abstract: | This study aims to classify high-growth firms using several machine learning algorithms, including K-Nearest Neighbors, Logistic Regression with L1 (Lasso) and L2 (Ridge) Regularization, XGBoost, Gradient Descent, Naive Bayes and Random Forest. Leveraging a dataset composed of financial metrics and firm characteristics between 2009 and 2022 with 1, 318, 799 unique firms (averaging 554, 178 annually), we evaluate the performance of each model using metrics such as MCC, ROC AUC, accuracy, precision, recall and F1-score. In our study, ROC AUC values ranged from 0.53 to 0.87 for employee-high growth and from 0.53 to 0.91 for turnover-high growth, depending on the method used. Our findings indicate that XGBoost achieves the highest performance, followed by Random Forest and Logistic Regression, demonstrating their effectiveness in distinguishing between high-growth and non-high-growth firms. Conversely, KNN and Naive Bayes yield lower accuracy. Furthermore, our findings reveal that growth opportunity emerges as the most significant factor in our study. This research contributes valuable insights in identifying high-growth firms and underscores the potential of machine learning in economic prediction. |
| Keywords: | High-growth firms, Machine learning, Prediction, Firm dynamics |
| JEL: | C40 C55 C60 C81 L25 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:tcb:wpaper:2413 |
| By: | Jule Schuettler (University of St.Gallen); Francesco Audrino (University of St. Gallen; Swiss Finance Institute); Fabio Sigrist (Lucerne University of Applied Sciences and Arts) |
| Abstract: | We present a novel approach to sentiment analysis in financial markets by using a state-of-the-art large language model, a market data-driven labeling approach, and a large dataset consisting of diverse financial text sources including earnings call transcripts, newspapers, and social media tweets. Based on our approach, we define a predictive high-low sentiment asset pricing factor which is significant in explaining cross-sectional asset pricing for U.S. stocks. Further, we find that a long/short equal-weighted portfolio yields an average annualized return of 35.56% and an annualized Sharpe ratio of 2.21, remaining substantially profitable even when transaction costs are considered. A comparison with an alternative financial sentiment analysis tool (FinBERT) underscores the superiority of our data-driven labeling approach over traditional human-annotated labeling. |
| Keywords: | natural language processing, large language models, DeBERTa, asset pricing |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:chf:rpseri:rp2469 |
| By: | Francesco Audrino (University of St. Gallen; Swiss Finance Institute); Jonathan Chassot (University of St. Gallen) |
| Abstract: | We investigate the predictive abilities of the heterogeneous autoregressive (HAR) model compared to machine learning (ML) techniques across an unprecedented dataset of 1, 445 stocks. Our analysis focuses on the role of fitting schemes, particularly the training window and re-estimation frequency, in determining the HAR model's performance. Despite extensive hyperparameter tuning, ML models fail to surpass the linear benchmark set by HAR when utilizing a refined fitting approach for the latter. Moreover, the simplicity of HAR allows for an interpretable model with drastically lower computational costs. We assess performance using QLIKE, MSE, and realized utility metrics, finding that HAR consistently outperforms its ML counterparts when both rely solely on realized volatility and VIX as predictors. Our results underscore the importance of a correctly specified fitting scheme. They suggest that properly fitted HAR models provide superior forecasting accuracy, establishing robust guidelines for their practical application and use as a benchmark. This study not only reaffirms the efficacy of the HAR model but also provides a critical perspective on the practical limitations of ML approaches in realized volatility forecasting. |
| Keywords: | Forecasting practice, HAR, Machine learning, Realized volatility, Volatility forecasting |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:chf:rpseri:rp2470 |
| By: | Masahiro Kato |
| Abstract: | This study proposes a debiasing method for smooth nonparametric estimators. While machine learning techniques such as random forests and neural networks have demonstrated strong predictive performance, their theoretical properties remain relatively underexplored. Specifically, many modern algorithms lack assurances of pointwise asymptotic normality and uniform convergence, which are critical for statistical inference and robustness under covariate shift and have been well-established for classical methods like Nadaraya-Watson regression. To address this, we introduce a model-free debiasing method that guarantees these properties for smooth estimators derived from any nonparametric regression approach. By adding a correction term that estimates the conditional expected residual of the original estimator, or equivalently, its estimation error, we obtain a debiased estimator with proven pointwise asymptotic normality, uniform convergence, and Gaussian process approximation. These properties enable statistical inference and enhance robustness to covariate shift, making the method broadly applicable to a wide range of nonparametric regression problems. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.20173 |
| By: | Kemal Kirtac; Guido Germano |
| Abstract: | We investigate the efficacy of large language models (LLMs) in sentiment analysis of U.S. financial news and their potential in predicting stock market returns. We analyze a dataset comprising 965, 375 news articles that span from January 1, 2010, to June 30, 2023; we focus on the performance of various LLMs, including BERT, OPT, FINBERT, and the traditional Loughran-McDonald dictionary model, which has been a dominant methodology in the finance literature. The study documents a significant association between LLM scores and subsequent daily stock returns. Specifically, OPT, which is a GPT-3 based LLM, shows the highest accuracy in sentiment prediction with an accuracy of 74.4%, slightly ahead of BERT (72.5%) and FINBERT (72.2%). In contrast, the Loughran-McDonald dictionary model demonstrates considerably lower effectiveness with only 50.1% accuracy. Regression analyses highlight a robust positive impact of OPT model scores on next-day stock returns, with coefficients of 0.274 and 0.254 in different model specifications. BERT and FINBERT also exhibit predictive relevance, though to a lesser extent. Notably, we do not observe a significant relationship between the Loughran-McDonald dictionary model scores and stock returns, challenging the efficacy of this traditional method in the current financial context. In portfolio performance, the long-short OPT strategy excels with a Sharpe ratio of 3.05, compared to 2.11 for BERT and 2.07 for FINBERT long-short strategies. Strategies based on the Loughran-McDonald dictionary yield the lowest Sharpe ratio of 1.23. Our findings emphasize the superior performance of advanced LLMs, especially OPT, in financial market prediction and portfolio management, marking a significant shift in the landscape of financial analysis tools with implications to financial regulation and policy analysis. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.19245 |
| By: | Denisa Millo; Blerina Vika; Nevila Baci |
| Abstract: | The financial sector, a pivotal force in economic development, increasingly uses the intelligent technologies such as natural language processing to enhance data processing and insight extraction. This research paper through a review process of the time span of 2018-2023 explores the use of text mining as natural language processing techniques in various components of the financial system including asset pricing, corporate finance, derivatives, risk management, and public finance and highlights the need to address the specific problems in the discussion section. We notice that most of the research materials combined probabilistic with vector-space models, and text-data with numerical ones. The most used technique regarding information processing is the information classification technique and the most used algorithms include the long-short term memory and bidirectional encoder models. The research noticed that new specific algorithms are developed and the focus of the financial system is mainly on asset pricing component. The research also proposes a path from engineering perspective for researchers who need to analyze financial text. The challenges regarding text mining perspective such as data quality, context-adaption and model interpretability need to be solved so to integrate advanced natural language processing models and techniques in enhancing financial analysis and prediction. Keywords: Financial System (FS), Natural Language Processing (NLP), Software and Text Engineering, Probabilistic, Vector-Space, Models, Techniques, TextData, Financial Analysis. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.20438 |
| By: | Vanessa Heinemann-Heile (Paderborn University) |
| Abstract: | I investigate whether a machine learning model can reliably predict firms’ tax rate perception. While standard models assume that decision-makers in firms are perfectly informed about firms’ tax rates and tax implications, also their tax rate perception influences the way in which they incorporate taxes into their decision-making processes. However, studies examining firms’ tax rate perception and its consequences remain scarce, mostly due to a lack of observations of firms’ tax rate perception. Using a dataset of German SMEs, I apply machine learning in the form of Extreme Gradient Boosting, to predict firms’ tax rate perception based on firm and personal characteristics of the decision-maker. The results show that Extreme Gradient Boosting outperforms traditional OLS regression. The model is highly accurate, as evidenced by a mean prediction error of less than one percentage point, produces reasonably precise predictions, as indicated by the root mean square error being comparable to the standard deviation, and explains up to 23.2% of the variance in firms’ tax rate perception. Even based on firm characteristics only, the model maintains high accuracy, albeit with some decline in precision and explained variance. Consistent with this finding, Shapley values highlight the importance of firm and personal characteristics such as tax compliance costs, tax literacy, and trust in government for the prediction. The results show that machine learning models can provide a time- and cost-effective way to fill the information gap created by the lack of observations on firms’ tax rate perception. This approach allows researchers and policymakers, to further analyze the impact of firms’ tax rate perception on tax reforms, tax compliance, or business decisions. |
| Keywords: | Tax Rate Perception, Business Taxation, Prediction, XGBoost, Shapley |
| JEL: | H25 D91 C8 C53 |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:pdn:dispap:128 |
| By: | Abdollah Rida |
| Abstract: | Credit Scoring is one of the problems banks and financial institutions have to solve on a daily basis. If the state-of-the-art research in Machine and Deep Learning for finance has reached interesting results about Credit Scoring models, usage of such models in a heavily regulated context such as the one in banks has never been done so far. Our work is thus a tentative to challenge the current regulatory status-quo and introduce new BASEL 2 and 3 compliant techniques, while still answering the Federal Reserve Bank and the European Central Bank requirements. With the help of Gradient Boosting Machines (mainly XGBoost) we challenge an actual model used by BANK A for scoring through the door Auto Loan applicants. We prove that the usage of such algorithms for Credit Scoring models drastically improves performance and default capture rate. Furthermore, we leverage the power of Shapley Values to prove that these relatively simple models are not as black-box as the current regulatory system thinks they are, and we attempt to explain the model outputs and Credit Scores within the BANK A Model Design and Validation framework |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.20225 |
| By: | Nico Herrig |
| Abstract: | This work aims to implement Long Short-Term Memory mixture density networks (LSTM-MDNs) for Value-at-Risk forecasting and compare their performance with established models (historical simulation, CMM, and GARCH) using a defined backtesting procedure. The focus was on the neural network's ability to capture volatility clustering and its real-world applicability. Three architectures were tested: a 2-component mixture density network, a regularized 2-component model (Arimond et al., 2020), and a 3-component mixture model, the latter being tested for the first time in Value-at-Risk forecasting. Backtesting was performed on three stock indices (FTSE 100, S&P 500, EURO STOXX 50) over two distinct two-year periods (2017-2018 as a calm period, 2021-2022 as turbulent). Model performance was assessed through unconditional coverage and independence assumption tests. The neural network's ability to handle volatility clustering was validated via correlation analysis and graphical evaluation. Results show limited success for the neural network approach. LSTM-MDNs performed poorly for 2017/2018 but outperformed benchmark models in 2021/2022. The LSTM mechanism allowed the neural network to capture volatility clustering similarly to GARCH models. However, several issues were identified: the need for proper model initialization and reliance on large datasets for effective learning. The findings suggest that while LSTM-MDNs provide adequate risk forecasts, further research and adjustments are necessary for stable performance. |
| Date: | 2025–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2501.01278 |
| By: | Trent Lockyer (Reserve Bank of New Zealand) |
| Abstract: | Key findings: • In this Note, we construct new high-frequency indicators which measure the sentiment of New Zealand news articles over time. • We test the usefulness of these news sentiment indicators as a measure of early-stage financial stress in New Zealand, one of several possible applications for these indicators. • The news sentiment indicators provide similar information to measures of consumer and business confidence. As the sentiment measures are updated more frequently than the survey-based confidence measures they can provide more timely information and more clearly identify the effect of specific events on consumer and business sentiment. • Our results suggest the news sentiment indicators can be a useful complement to the forward-looking indicators of financial stress we monitor, and we are considering how to enhance and build these techniques into our ongoing assessment of financial stability. |
| Date: | 2024–07 |
| URL: | https://d.repec.org/n?u=RePEc:nzb:nzbans:2024/07 |
| By: | Cormun, Vito; Ristolainen, Kim |
| Abstract: | Leveraging Wall Street Journal news, recent developments in textual analysis, and generative AI, we estimate a narrative decomposition of the dollar exchange rate. Our findings shed light on the connection between economic fundamentals and the exchange rate, as well as on its absence. From the late 1970s onwards, we identify six distinct narratives that explain changes in the exchange rate, each largely non-overlapping. U.S. fiscal and monetary policies play a significant role in the early part of the sample, while financial market news becomes more dominant in the second half. Notably, news on technological change predicts the exchange rate throughout the entire sample period. Finally, using text-augmented regressions, we find evidence that media coverage explains the unstable relationship between exchange rates and macroeconomic indicators. |
| Keywords: | Exchange rates, big data, textual analysis, macroeconomic news, Wall Street Journal, narrative retrieval, scapegoat |
| JEL: | C3 C5 F3 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:bofrdp:306349 |
| By: | Francesco Audrino (University of St. Gallen; Swiss Finance Institute); Jessica Gentner (University of St. Gallen; Swiss National Bank); Simon Stalder (Swiss National Bank; University of Lugano) |
| Abstract: | This paper presents an innovative method for measuring uncertainty using Large Language Models (LLMs), offering enhanced precision and contextual sensitivity compared to the conventional methods used to construct prominent uncertainty indices. By analyzing newspaper texts with state-of-the-art LLMs, our approach captures nuances often missed by conventional methods. We develop indices for various types of uncertainty, including geopolitical risk, economic policy, monetary policy, and financial market uncertainty. Our findings show that shocks to these LLM-based indices exhibit stronger associations with macroeconomic variables, shifts in investor behaviour, and asset return variations than conventional indices, underscoring their potential for more accurately reflecting uncertainty. |
| Keywords: | Large Language Models, Economic policy, Geopolitical risk, Monetary policy, Financial markets, Uncertainty measurment |
| JEL: | C45 C55 E44 G12 |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:chf:rpseri:rp2468 |
| By: | Yichen Luo; Yebo Feng; Jiahua Xu; Paolo Tasca; Yang Liu |
| Abstract: | Cryptocurrency investment is inherently difficult due to its shorter history compared to traditional assets, the need to integrate vast amounts of data from various modalities, and the requirement for complex reasoning. While deep learning approaches have been applied to address these challenges, their black-box nature raises concerns about trust and explainability. Recently, large language models (LLMs) have shown promise in financial applications due to their ability to understand multi-modal data and generate explainable decisions. However, single LLM faces limitations in complex, comprehensive tasks such as asset investment. These limitations are even more pronounced in cryptocurrency investment, where LLMs have less domain-specific knowledge in their training corpora. To overcome these challenges, we propose an explainable, multi-modal, multi-agent framework for cryptocurrency investment. Our framework uses specialized agents that collaborate within and across teams to handle subtasks such as data analysis, literature integration, and investment decision-making for the top 30 cryptocurrencies by market capitalization. The expert training module fine-tunes agents using multi-modal historical data and professional investment literature, while the multi-agent investment module employs real-time data to make informed cryptocurrency investment decisions. Unique intrateam and interteam collaboration mechanisms enhance prediction accuracy by adjusting final predictions based on confidence levels within agent teams and facilitating information sharing between teams. Empirical evaluation using data from November 2023 to September 2024 demonstrates that our framework outperforms single-agent models and market benchmarks in classification, asset pricing, portfolio, and explainability performance. |
| Date: | 2025–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2501.00826 |
| By: | Sarr, Ibrahima; Dang, Hai-Anh H.; Guzman Gutierrez, Carlos Santiago; Beltramo, Theresa; Verme, Paolo |
| Abstract: | Household consumption or income surveys do not typically cover refugee populations. In the rare cases where refugees are included, inconsistencies between different data sources could interfere with comparable poverty estimates. We test the performance of a recently developed cross-survey imputation method to estimate poverty for a sample of refugees in Colombia, combining household income surveys collected by the Government of Colombia and administrative (ProGres) data collected by the United Nations High Commissioner for Refugees in 2019 and 2022. We find that certain variable transformation methods can help resolve these inconsistencies. Estimation results with our preferred variable standardization method are robust to different imputation methods, including the normal linear regression method, the empirical distribution of the errors method, and the probit and logit methods. Several common machine learning techniques generally perform worse than our proposed imputation methods. We also find that we can reasonably impute poverty rates using an older household income survey and a more recent ProGres dataset for most of the poverty lines. These results provide relevant inputs into designing better surveys and administrative datasets on refugees in various country settings. |
| Keywords: | refugees, poverty, imputation, Colombia |
| JEL: | C15 F22 I32 O15 O20 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:glodps:1534 |
| By: | Wladislaw Mill; Tobias Ebert; Jana B. Berkessel; Thorsteinn Jonsson; Sune Lehmann; Jochen E. Gebauer |
| Abstract: | Does war make people more religious? Answers to this classic question are dominated by the lack of causality. We exploit the Vietnam Draft Lottery -- a natural experiment that drafted male U.S. citizens into military service during the Vietnam War -- to conclusively show that war increases religiosity. We measure religiosity via religious imagery on web-scraped photographs of hundreds of thousands of gravestones of deceased U.S. Americans using a tailor-made convolutional neural network. Our analysis provides compelling and robust evidence that war indeed increases religiosity: people who were randomly drafted into war are at least 20% more likely to have religious gravestones. This effect sets in almost immediately, persists even after 50 years, and generalizes across space and societal strata. |
| Keywords: | War, Religion, Vietnam Draft Lottery, Grave |
| JEL: | Z12 N30 N40 P00 |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:bon:boncrc:crctr224_2024_614 |
| By: | Thomas Gaertner; Christoph Lippert; Stefan Konigorski |
| Abstract: | In response to the growing demand for accurate demand forecasts, this research proposes a generalized automated sales forecasting pipeline tailored for small- to medium-sized enterprises (SMEs). Unlike large corporations with dedicated data scientists for sales forecasting, SMEs often lack such resources. To address this, we developed a comprehensive forecasting pipeline that automates time series sales forecasting, encompassing data preparation, model training, and selection based on validation results. The development included two main components: model preselection and the forecasting pipeline. In the first phase, state-of-the-art methods were evaluated on a showcase dataset, leading to the selection of ARIMA, SARIMAX, Holt-Winters Exponential Smoothing, Regression Tree, Dilated Convolutional Neural Networks, and Generalized Additive Models. An ensemble prediction of these models was also included. Long-Short-Term Memory (LSTM) networks were excluded due to suboptimal prediction accuracy, and Facebook Prophet was omitted for compatibility reasons. In the second phase, the proposed forecasting pipeline was tested with SMEs in the food and electric industries, revealing variable model performance across different companies. While one project-based company derived no benefit, others achieved superior forecasts compared to naive estimators. Our findings suggest that no single model is universally superior. Instead, a diverse set of models, when integrated within an automated validation framework, can significantly enhance forecasting accuracy for SMEs. These results emphasize the importance of model diversity and automated validation in addressing the unique needs of each business. This research contributes to the field by providing SMEs access to state-of-the-art sales forecasting tools, enabling data-driven decision-making and improving operational efficiency. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.20420 |
| By: | Philipp Bach; Victor Chernozhukov; Sven Klaassen; Martin Spindler; Jan Teichert-Kluge; Suhas Vijaykumar |
| Abstract: | This paper advances empirical demand analysis by integrating multimodal product representations derived from artificial intelligence (AI). Using a detailed dataset of toy cars on \textit{Amazon.com}, we combine text descriptions, images, and tabular covariates to represent each product using transformer-based embedding models. These embeddings capture nuanced attributes, such as quality, branding, and visual characteristics, that traditional methods often struggle to summarize. Moreover, we fine-tune these embeddings for causal inference tasks. We show that the resulting embeddings substantially improve the predictive accuracy of sales ranks and prices and that they lead to more credible causal estimates of price elasticity. Notably, we uncover strong heterogeneity in price elasticity driven by these product-specific features. Our findings illustrate that AI-driven representations can enrich and modernize empirical demand analysis. The insights generated may also prove valuable for applied causal inference more broadly. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2501.00382 |
| By: | Yoosoon Chang (Department of Economics, Indiana University and Centre for Applied Macroeconomics and Commodity Prices (CAMP) at the BI Norwegian Business School); Steven N. Durlauf (University of Chicago); Bo Hu (Department of Economics, Indiana University); Joon K. Park (Department of Economics, Indiana University) |
| Abstract: | This paper proposes a fully nonparametric model to investigate the dynamics of intergenerational income mobility for discrete outcomes. In our model, an individual’s income class probabilities depend on parental income in a manner that accommodates nonlinearities and interactions among various individual and parental characteristics, including race, education, and parental age at childbearing. Consequently, we offer a generalization of Markov chain mobility models. We employ kernel techniques from machine learning and further regularization for estimating this highly flexible model. Utilizing data from the Panel Study of Income Dynamics (PSID), we find that race and parental education play significant roles in determining the influence of parental income on children’s economic prospects. |
| Keywords: | intergenerational income mobility, ordered multinomial probability model, nonparametric estimation, heterogeneous treatment effects, reproducing kernel Hilbert space, effects of parental education |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:inu:caeprp:2024008 |
| By: | Filip Stefaniuk (University of Warsaw, Faculty of Economic Sciences); Robert Ślepaczuk (University of Warsaw, Faculty of Economic Sciences, Department of Quantitative Finance and Machine Learning, Quantitative Finance Research Group) |
| Abstract: | The thesis investigates the usage of Informer architecture for building automated trading strategies for high frequency Bitcoin data. Two strategies using Informer models with different loss functions, Quantile loss and Generalized Mean Absolute Directional Loss (GMADL), are proposed and evaluated against the Buy and Hold benchmark and two benchmark strategies based on technical indicators. The evaluation is conducted using data of various frequencies: 5 minute, 15 minute, and 30 minute intervals, over the 6 different periods. Although the Informer-based model with Quantile loss did not manage to outperform the benchmark, the model that uses novel GMADL loss function turned out to be benefiting from higher frequency data and beat all the other strategies on most of the testing periods. The primary contribution of this study is the application and assessment of the Quantile and GMADL loss functions with the Informer model to forecast future returns, subsequently using these forecasts to develop automated trading strategies. The research provides evidence that employing an Informer model trained with the GMADL loss function can result in superior trading outcomes compared to the buy-and-hold approach. |
| Keywords: | Machine Learning, Financial Series Forecasting, Automated Trading Strategy, Informer, Transformer, Bitcoin, High Frequency Trading, Statistics, GMADL |
| JEL: | C4 C14 C45 C53 C58 G13 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:war:wpaper:2024-27 |
| By: | Hannes Mueller; Christopher Rauh; Benjamin R Seimon; Mr. Raphael A Espinoza |
| Abstract: | Can macroeconomic policy effectively help prevent armed conflicts? This paper contends that two key criteria need to be satisfied: the long-term benefits of prevention policies must exceed the costs associated with uncertain forecasts, and the policies themselves must be directly able to contribute to conflict prevention. This paper proposes policy simulations, based on a novel method of Mueller et al (2024a) that integrates machine learning and dynamic optimization, to show that investing in prevention can generate huge long-run benefits. Returns to prevention policies in countries that have not suffered recently from violence range from $26 to $75 per $1 spent on prevention, and for countries with recent violence, the rate of return could be as high as $103 per $1 spent on prevention. Furthermore, an analysis of the available data and results in the literature suggest that sound macroeconomic policies and international support for these policies can play key roles in conflict prevention. Based on these findings, this paper proposes actionable recommendations, for both global and domestic policymakers as well as international financial institutions and multilateral organizations, to promote peace and stability through macroeconomic policy. |
| Keywords: | Prevention; fragile state; conflict ; machine learning |
| Date: | 2024–12–20 |
| URL: | https://d.repec.org/n?u=RePEc:imf:imfwpa:2024/256 |
| By: | ITO Arata; SATO Masahiro; OTA Rui |
| Abstract: | Policy uncertainty has the potential to reduce policy effectiveness. Existing studies have measured policy uncertainty by tracking the frequency of specific keywords in newspaper articles. However, this keyword-based approach fails to account for the context of the articles and differentiate the types of uncertainty that such contexts indicate. This study introduces a new method of measuring different types of policy uncertainty in news content which utilizes large language models (LLMs). Specifically, we differentiate policy uncertainty into forward-looking and backward-looking uncertainty, or in other words, uncertainty regarding future policy direction and uncertainty about the effectiveness of the current policy. We fine-tune the LLMs to identify each type of uncertainty expressed in newspaper articles based on their context, even in the absence of specific keywords indicating uncertainty. By applying this method, we measure Japan’s monetary policy uncertainty (MPU) from 2015 to 2016. To reflect the unprecedented monetary policy conditions during this period when the unconventional policies were taken, we further classify MPU by layers of policy changes: changes in specific market operations and changes in the broader policy framework. The experimental results show that our approach successfully captures the dynamics of MPU, particularly for forward-looking uncertainty, which is not fully captured by the existing approach. Forward- and backward-looking uncertainty indices exhibit distinct movements depending on the conditions under which changes in the policy framework occur. This suggests that perceived uncertainty regarding monetary policy would be state-dependent, varying with the prevailing social environment. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:eti:dpaper:24080 |
| By: | Fadavi, Sara; Hillert, Alexander |
| Abstract: | SAFE's monthly Manager Sentiment Index is constructed by extracting sentiment from corporate financial disclosures of listed companies in Germany, offering significant insights into top management's perspectives. This white paper outlines the methodology behind the index and its financial implications. Information about managers' assessment of firms' performance and financial conditions is material to investors but, at the same time, hard to observe. The SAFE Manager Sentiment Index quantifies managers' beliefs using textual analysis of financial reports and earnings conference call transcripts. We show that the index is a strong predictor of future stock market returns. In summary, the SAFE Manager Sentiment Index provides a practical tool for key stakeholders such as investors, analysts, and policymakers seeking timely signals of corporate sentiment. |
| Keywords: | Manager Sentiment, Textual Analysis, Financial Disclosures, Return Predictability |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:safewh:308085 |
| By: | Yijia Xiao; Edward Sun; Di Luo; Wei Wang |
| Abstract: | Significant progress has been made in automated problem-solving using societies of agents powered by large language models (LLMs). In finance, efforts have largely focused on single-agent systems handling specific tasks or multi-agent frameworks independently gathering data. However, multi-agent systems' potential to replicate real-world trading firms' collaborative dynamics remains underexplored. TradingAgents proposes a novel stock trading framework inspired by trading firms, featuring LLM-powered agents in specialized roles such as fundamental analysts, sentiment analysts, technical analysts, and traders with varied risk profiles. The framework includes Bull and Bear researcher agents assessing market conditions, a risk management team monitoring exposure, and traders synthesizing insights from debates and historical data to make informed decisions. By simulating a dynamic, collaborative trading environment, this framework aims to improve trading performance. Detailed architecture and extensive experiments reveal its superiority over baseline models, with notable improvements in cumulative returns, Sharpe ratio, and maximum drawdown, highlighting the potential of multi-agent LLM frameworks in financial trading. |
| Date: | 2024–12 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2412.20138 |
| By: | Fiorella De Fiore; Alexis Maurin; Andrej Mijakovic; Damiano Sandri |
| Abstract: | We analyse the media's role in channelling information about the Fed's monetary policy stance to the public. Using LLMs, we find a tight correspondence between FOMC communication and media coverage, although with significant variation over time. The communication pass-through weakened during the ZLB period and improved with the introduction of press conferences, which now exert strong influence on the media. Media coverage effects households' inflation expectations, particularly when inflation is high and volatile, while we do not detect a direct impact of FOMC communication. This underscores the media's crucial function in channelling central banks' communication to the public. |
| Keywords: | central bank communication, media coverage, large language models, households' expectations |
| JEL: | E50 E52 E58 |
| URL: | https://d.repec.org/n?u=RePEc:bis:biswps:1231 |