-
Towards reducing hallucination in extracting information from financial reports using Large Language Models
Authors:
Bhaskarjit Sarmah,
Tianjie Zhu,
Dhagash Mehta,
Stefano Pasquali
Abstract:
For a financial analyst, the question and answer (Q\&A) segment of the company financial report is a crucial piece of information for various analysis and investment decisions. However, extracting valuable insights from the Q\&A section has posed considerable challenges as the conventional methods such as detailed reading and note-taking lack scalability and are susceptible to human errors, and Op…
▽ More
For a financial analyst, the question and answer (Q\&A) segment of the company financial report is a crucial piece of information for various analysis and investment decisions. However, extracting valuable insights from the Q\&A section has posed considerable challenges as the conventional methods such as detailed reading and note-taking lack scalability and are susceptible to human errors, and Optical Character Recognition (OCR) and similar techniques encounter difficulties in accurately processing unstructured transcript text, often missing subtle linguistic nuances that drive investor decisions. Here, we demonstrate the utilization of Large Language Models (LLMs) to efficiently and rapidly extract information from earnings report transcripts while ensuring high accuracy transforming the extraction process as well as reducing hallucination by combining retrieval-augmented generation technique as well as metadata. We evaluate the outcomes of various LLMs with and without using our proposed approach based on various objective metrics for evaluating Q\&A systems, and empirically demonstrate superiority of our method.
△ Less
Submitted 16 October, 2023;
originally announced October 2023.
-
Risk Management and Return Prediction
Authors:
Qingyin Ge,
Yunuo Ma,
Yuezhi Liao,
Rongyu Li,
Tianle Zhu
Abstract:
With the good development in the financial industry, the market starts to catch people's eyes, not only by the diversified investing choices ranging from bonds and stocks to futures and options but also by the general "high-risk, high-reward" mindset prompting people to put money in the financial market. People are interested in reducing risk at a given level of return since there is no way of hav…
▽ More
With the good development in the financial industry, the market starts to catch people's eyes, not only by the diversified investing choices ranging from bonds and stocks to futures and options but also by the general "high-risk, high-reward" mindset prompting people to put money in the financial market. People are interested in reducing risk at a given level of return since there is no way of having both high returns and low risk. Many researchers have been studying this issue, and the most pioneering one is Harry Markowitz's Modern Portfolio Theory developed in 1952, which is the cornerstone of investment portfolio management and aims at "maximum the return at the given risk". In contrast to that, fifty years later, E. Robert Fernholz's Stochastic Portfolio Theory, as opposed to the normative assumption served as the basis of earlier modern portfolio theory, is consistent with the observable characteristics of actual portfolios and markets. In this paper, after introducing some basic theories of Markowitz's MPT and Fernholz's SPT, then we step across to the application side, trying to figure out under four basic models based on Markowitz Efficient Frontier, including Markowitz Model, Constant Correlation Model, Single Index Model, and Multi-Factor Model, which portfolios will be selected and how do these portfolios perform in the real world. Here we also involve universal Portfolio Algorithmby Thomas M. Cover to select portfolios as a comparison. Besides, each portfolio value at Risk, Expected Shortfall, and corresponding bootstrap confidence interval for risk management will be evaluated. Finally, by utilizing factor analysis and time series models, we could predict the future performance of our four models.
△ Less
Submitted 28 June, 2020;
originally announced July 2020.
-
Benford's law first significant digit and distribution distances for testing the reliability of financial reports in developing countries
Authors:
Jing Shi,
Marcel Ausloos,
Tingting Zhu
Abstract:
We discuss a common suspicion about reported financial data, in 10 industrial sectors of the 6 so called "main developing countries" over the time interval [2000-2014]. These data are examined through Benford's law first significant digit and through distribution distances tests. It is shown that several visually anomalous data have to be a priori removed. Thereafter, the distributions much better…
▽ More
We discuss a common suspicion about reported financial data, in 10 industrial sectors of the 6 so called "main developing countries" over the time interval [2000-2014]. These data are examined through Benford's law first significant digit and through distribution distances tests. It is shown that several visually anomalous data have to be a priori removed. Thereafter, the distributions much better follow the first digit significant law, indicating the usefulness of a Benford's law test from the research starting line. The same holds true for distance tests. A few outliers are pointed out.
△ Less
Submitted 30 November, 2017;
originally announced December 2017.
