Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2611.04119

Multidimensional Views Behind Target-Price Revisions: An LLM-Based Representation of Analyst Outlooks

Kiyoun Kim·Sung-Yeon Nathan Lim
View original source ↗Read source document ↗Export metadata ↓

Abstract

A target price compresses an analyst's multidimensional views into a single number. We propose an LLM-based text representation that extracts these views from research reports as question-based outlook scores with supporting evidence. The questions organize reported views on earnings, valuation, industry conditions, and other investment dimensions into a compact representation of analyst sentiment. In walk-forward evaluation, the scores improve prediction of subsequent target-price revisions beyond current-revision momentum and sector differences. They outperform financial dictionary and FinBERT sentiment measures. These findings suggest that structured questions can capture information about subsequent analyst actions that is not fully reflected in the current numerical revision. By linking each dimension to an explicit investment question and supporting report evidence, the approach makes the extracted outlooks available for inspection and interpretation. It offers a way to complement target-price signals with a multidimensional view of the judgments expressed in analyst narratives.

Explore related subjects

q-fin.CP

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Kiyoun Kim, Sung-Yeon Nathan Lim. 2026-11-03. Multidimensional Views Behind Target-Price Revisions: An LLM-Based Representation of Analyst Outlooks. https://arxiv.org/abs/2611.04119

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

CosmosAlpha: Multi-Relational Graph Search with Pareto-Guided Exploration for LLM-Driven Alpha Mining

Discovering predictive alpha factors from noisy financial data is a fundamental yet challenging problem in quantitative finance. LLM-driven formulaic alpha mining offers an attractive paradigm by combining interpretable symbolic expressions with iterative generation and refinement guided by quantitative feedback. However, existing generation-centric and tree-based search strategies primarily organize factors through generation lineage, leaving structural and statistical relationships across search branches underutilized. Consequently, information remains localized along parent--child trajectories, which can confine exploration to redundant regions. Moreover, effective factor discovery must account for multiple distinct objectives that cannot be adequately captured by a single search criterion. We introduce CosmosAlpha, a graph-based LLM factor-mining framework that represents discovered factors as a multi-relational graph encoding generation lineage, expression-structure similarity, and cross-sectional exposure correlation. A Graph Flow mechanism propagates search utilities across the graph, allowing each factor to be assessed using relational evidence from nearby factors rather than in isolation. Pareto-guided selection jointly considers quality, fertility and diversity, explicitly modeling their trade-offs within a multi-objective search framework. Across CSI 300 and CSI 500, CosmosAlpha-Graph achieves the best mean rank over seven predictive and portfolio metrics and outperforms its controlled CosmosAlpha-MCTS counterpart on every reported metric, showing a consistent advantage for relational graph search in LLM-driven factor mining. The source code is available at https://github.com/wzhmtldswz/CosmosAlpha

q-fin.CP↗
arXiv2026

Efficient Monte Carlo Valuation of Corporate Bonds in Financial Networks

Valuing corporate bonds in systemic economies is challenging due to intricate webs of inter-institutional exposures. When a bank defaults, cascading losses propagate through the network, with payments determined by a system of fixed-point equations lacking closed-form solutions. Standard Monte Carlo methods cannot capture rare yet critical default events, while existing rare-event simulation techniques fail to account for higher-order network effects and scale poorly with network size. To overcome these challenges, we propose a novel approach -- Bi-Level Importance Sampling with Splitting -- and characterize individual bank defaults by decoupling them from the network's complex fixed-point dynamics. This separation enables a two-stage estimation process that directly generates samples from the banks' default events. We demonstrate theoretically that the method is both scalable and asymptotically optimal, and validate its effectiveness through numerical studies on both synthetic and empirically observed networks.

q-fin.CP↗
arXiv2026

A New GAUSS Software Component for Implementing Dynamic Tests of Contagion with an Application

The main goal of this article is to present a new software component created by the authors in GAUSS for implementing dynamic tests for contagion. These tests are aimed at providing the possibilities of testing for spiller-over effects between financial markets during a crisis or during a pandemic situation such as the one that we experienced recently. Unlike standard tests for contagion that have fixed dates for a potential shift, the tests implemented here are dynamic in the sense that the potential contagion effect is measured across the time span. The leverage bootstrap simulation technique is utilized in order to generate critical values that are reliable when the tests are implemented in order to make sure that the conducted inference is accurate. Since the approach is new, software component needs to be created for conducting the underlying tests, which is accomplished via this article. The module is very easy to implement by the general practitioners via a graphical user interface. An application is offered by using real data from stock markets in order to demonstrate how the software component can be used.

q-fin.CP↗