BlackRock (United States)

Recent advancements in Generative AI, such as Large Language Models (LLMs), have demonstrated remarkable success across various general tasks. Extensive studies have explored leveraging generative models in finance, but significant challenges persist. This half-day workshop explores potential approaches and research directions to address these challenges by equipping generative models with advanced Information Retrieval (IR) models. Specifically, this workshop seeks to provide a platform for discussing innovative ideas that facilitate the advancement of IR technology to enrich generative models in finance from four key perspectives: (i) financial IR techniques (ii) financial IR benchmarking and evaluation (iii) financial systems and agents/assistants (iv) and trustworthiness, privacy and security when applying financial IR and generative models. This workshop aims to deepen understanding, accelerate progress, and support the advancement of IR technology to enhance generative models to address financial challenges.

Climate change threatens global food systems1, but the extent to which adaptation will reduce losses remains unknown and controversial2. Even within the well-studied context of US agriculture, some analyses argue that adaptation will be widespread and climate damages small3,4, whereas others conclude that adaptation will be limited and losses severe5,6. Scenario-based analyses indicate that adaptation should have notable consequences on global agricultural productivity7, 8–9, but there has been no systematic study of how extensively real-world producers actually adapt at the global scale. Here we empirically estimate the impact of global producer adaptations using longitudinal data on six staple crops spanning 12,658 regions, capturing two-thirds of global crop calories. We estimate that global production declines 5.5 × 1014 kcal annually per 1 °C global mean surface temperature (GMST) rise (120 kcal per person per day or 4.4% of recommended consumption per 1 °C; P < 0.001). We project that adaptation and income growth alleviate 23% of global losses in 2050 and 34% at the end of the century (6% and 12%, respectively; moderate-emissions scenario), but substantial residual losses remain for all staples except rice. In contrast to analyses of other outcomes that project the greatest damages to the global poor10,11, we find that global impacts are dominated by losses to modern-day breadbaskets with favourable climates and limited present adaptation, although losses in low-income regions losses are also substantial. These results indicate a scale of innovation, cropland expansion or further adaptation that might be necessary to ensure food security in a changing climate.

In many real-world classification or recognition tasks, it is often difficult to collect training examples that exhaust all possible classes due to, for example, incomplete knowledge during training or ever changing regimes. Therefore, samples from unknown/novel classes may be encountered in testing/deployment. In such scenarios, the classifiers should be able to i) perform classification on known classes, and at the same time, ii) identify samples from unknown classes. This is known as open-set recognition. Although random forest has been an extremely successful framework as a generalpurpose classification (and regression) method, in practice, it usually operates under the closed-set assumption and is not able to identify samples from new classes when run out of the box. In this work, we propose a novel approach to enabling open-set recognition capability for random forest classifiers by incorporating distance metric learning and distance-based open-set recognition. The proposed method is validated on both synthetic and real-world datasets. The experimental results indicate that the proposed approach outperforms state-of-the-art distance-based open-set recognition methods.

Understanding non-linear relationships among financial instruments has various applications in investment processes ranging from risk management, portfolio construction and trading strategies. Here, we focus on interconnectedness among stocks based on their correlation matrix which we represent as a network with the nodes representing individual stocks and the weighted links between pairs of nodes representing the corresponding pair-wise correlation coefficients. The traditional network science techniques, which are extensively utilized in financial literature, require handcrafted features such as centrality measures to understand such correlation networks. However, manually enlisting all such handcrafted features may quickly turn out to be a daunting task. Instead, we propose a new approach for studying nuances and relationships within the correlation network in an algorithmic way using a graph machine learning algorithm called Node2Vec. In particular, the algorithm compresses the network into a lower dimensional continuous space, called an embedding, where pairs of nodes that are identified as similar by the algorithm are placed closer to each other. By using log returns of S&P 500 stock data, we show that our proposed algorithm can learn such an embedding from its correlation network. We define various domain specific quantitative (and objective) and qualitative metrics that are inspired by metrics used in the field of Natural Language Processing (NLP) to evaluate the embeddings in order to identify the optimal one. Further, we discuss various applications of the embeddings in investment management.

We analyze the effect of leverage on tax-managed public equity strategies for different types of investors by running back-tests on hypothetical portfolios. We found the average 130/30 long/short portfolio funded with cash generates 2.7 times more capital losses than the average long-only portfolio over the first 10 years. For an investor with a sufficient supply of short- and long-term gains, these additional losses translate to average pre-liquidation tax alpha (“TA”) of 4.41% per year. Liquidation reduces TA by about half, while a lack of short-term capital gains to offset reduces TA by approximately 40%. In combination, those investor characteristics may render a long-only loss harvesting portfolio or a low-cost ETF more desirable than a long/short tax-managed equity portfolio after costs and fees are taken into consideration. Leverage can also be used to revitalize an ossified portfolio whose loss-generating capacity has diminished.

A portfolio can be viewed as the collection of the businesses, policies and practices of constituent companies. We measure investors’ Ownership of this collection. Ownership metrics aggregate an assortment of company specific Environmental, Social and Governance (ESG) characteristics to the portfolio level, and they can inform investment and engagement decisions. Relative to a benchmark, investor Ownership is active and satisfies a zero-sum property, which underscores the distinction between Ownership and impact. Ownership of ESG characteristics may be interpreted as ascribing ethical responsibility, but that conclusion and any decisions that result from it belong to the investor.

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.