傳統金融文本情緒分析多依詞典法,於萃取階段即將新聞壓縮為單一極性維度。本研究以兩個獨立開源之大型語言模型(LargeLanguage Models, LLMs;Gemma 4 E4B 與 Llama 3.1 8B), 對 2015–2024 年約 250萬篇S&P500新聞萃取三維情緒評分(極性、強度、不確定性),合成三項市場層級之新聞資訊結構(NewsInformationStructure, NIS) 指標 : 強度面向NISmagnitude 與離散面向NISdispersion 由三維評分聚合, 冗餘面向NISredundancy 則基於資料庫事件相似度、不依賴LLM評分。
依Bekaert and Hoerova (2014) 之 VIX 兩成分分解框架, 本研究檢驗NIS 指標對已實現波動率(RV)與變異風險溢酬(VRP)於週、月、季三個尺度之差別預測力,並以同期vs預測、real-timeregime條件性與非對稱反應等檢驗確認結構穩定性。實證結果顯示具區別性之指標分工:NISmagnitude 對 RV 與 VRP 呈現符號相反之預測力(對RV為負、對VRP 為正);NISredundancy 則主要透過 VRP 路徑作用, 於月度至季度尺度達峰;核心機制於兩個獨立LLM下方向一致。本研究將LLM多維情緒分解與VIX兩成分框架整合, 展示三維情緒評分可於市場層級呈現單一極性指標難以觀察之差別預測結構, 為文本驅動金融學提供多維測量之分析範例。

Traditional financial sentiment analysis, typically lexicon-based, compresses news semantics into a single polarity dimension at the extraction stage, yielding output that captures sentiment direction only. This study employs two independent open-source LLMs (Gemma 4 E4B and Llama 3.1 8B) to extract three-dimensional sentiment scores (polarity, intensity, uncertainty) from approximately 2.5 million S&P 500 news articles during 2015– 2024, and synthesizes three market-level News Information Structure (NIS) indicators: NISmagnitude (intensity), NISdispersion (dispersion), and NISredundancy (redundancy).
Following Bekaert and Hoerova’s (2014) VIX two-component decomposition framework, this study examines the differential predictive power of NIS indicators on realized volatility (RV) and the variance risk premium (VRP) across weekly, monthly, and quarterly horizons, with two robustness checks (contemporaneous-versus-predictive comparison, real-time regime conditionality) and one extension check (asymmetric response). Empirical results show a discernible indicator division: NISmagnitude exhibits opposite-signed predictive power on RV and VRP (negative on RV, positive on VRP); NISredundancy primarily predicts via the VRP channel and peaks at monthly-to-quarterly horizons; the core mechanism is directionally consistent across the two LLMs. This study integrates LLM-based multidimensional sentiment decomposition with the VIX two-component framework into a unified computable structure, demonstrating that three-dimensional sentiment scores can reveal, at the market level, a differential predictive structure not observable under a singlepolarity indicator, and thereby offering an example of multidimensional measurement for text-driven finance.

Andrews, D. W. K. (1991). Heteroskedasticity and autocorrelation consistent covariance matrix estimation. Econometrica, 59(3):817–858.
Antweiler, W. and Frank, M. Z. (2004). Is all that talk just noise? The information content of Internet stock message boards. Journal of Finance, 59(3):1259–1294.
Araci, D. (2019). FinBERT: Financial sentiment analysis with pre-trained language models. arXiv preprint arXiv:1908.10063.
Baker, M. and Wurgler, J. (2006). Investor sentiment and the cross-section of stock returns. Journal of Finance, 61(4):1645–1680.
Baker, S. R., Bloom, N., and Davis, S. J. (2016). Measuring economic policy uncertainty. Quarterly Journal of Economics, 131(4):1593–1636.
Bekaert, G. and Hoerova, M. (2014). The VIX, the variance premium and stock market volatility. Journal of Econometrics, 183(2):181–192.
Black, F. (1976). Studies of stock price volatility changes. In Proceedings of the 1976 Meeting of the Business and Economic Statistics Section, pages 177–181. American Statistical Association.
Bollerslev, T., Litvinova, J., and Tauchen, G. (2006). Leverage and volatility feedback effects in high-frequency data. Journal of Financial Econometrics, 4(3):353–384.
Bollerslev, T., Marrone, J., Xu, L., and Zhou, H. (2014). Stock return predictability and variance risk premia: statistical inference and international evidence. Journal of Financial and Quantitative Analysis, 49(3):633–661.
Bollerslev, T., Tauchen, G., and Zhou, H. (2009). Expected stock returns and variance risk premia. Review of Financial Studies, 22(11):4463–4492.
Christie, A. A. (1982). The stochastic behavior of common stock variances: Value, leverage and interest rate effects. Journal of Financial Economics, 10(4):407–432.
Clark, T. E. and West, K. D. (2007). Approximately normal tests for equal predictive accuracy in nested models. Journal of Econometrics, 138(1):291–311.
Cookson, J. A. and Niessner, M. (2020). Why don't we agree? Evidence from a social network of investors. Journal of Finance, 75(1):173–228.
Da, Z., Engelberg, J., and Gao, P. (2011). In search of attention. Journal of Finance, 66(5):1461–1499.
Drechsler, I. (2013). Uncertainty, time-varying fear, and asset prices. Journal of Finance, 68(5):1843–1889.
Epstein, L. G. and Schneider, M. (2008). Ambiguity, information quality, and asset pricing. Journal of Finance, 63(1):197–228.
García, D. (2013). Sentiment during recessions. Journal of Finance, 68(3):1267–1300.
Glasserman, P., Mamaysky, H., and Qin, J. (2023). New news is bad news. arXiv preprint arXiv:2309.05560.
Granger, C. W. J. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica, 37(3):424–438.
Hirshleifer, D. and Teoh, S. H. (2003). Limited attention, information disclosure, and financial reporting. Journal of Accounting and Economics, 36(1–3):337–386.
Koenker, R. and Bassett, G. (1978). Regression quantiles. Econometrica, 46(1):33–50.
Lopez-Lira, A. and Tang, Y. (2023). Can ChatGPT forecast stock price movements? Return predictability and large language models. SSRN Working Paper.
Loughran, T. and McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66(1):35–65.
Manela, A. and Moreira, A. (2017). News implied volatility and disaster concerns. Journal of Financial Economics, 123(1):137–162.
Moreira, A. and Muir, T. (2017). Volatility-managed portfolios. Journal of Finance, 72(4):1611–1644.
Newey, W. K. and West, K. D. (1987). A simple, positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica, 55(3):703–708.
Tetlock, P. C. (2007). Giving content to investor sentiment: The role of media in the stock market. Journal of Finance, 62(3):1139–1168.
Tetlock, P. C. (2011). All the news that's fit to reprint: Do investors react to stale information? Review of Financial Studies, 24(5):1481–1512.