Real-Time Stock Market Sentiment Analysis and Price Prediction Using Cloud-Native Architecture and Generative AI

Abstract

The rapid proliferation of digital financial news has created an unprecedented opportunity to harness unstructured textual data for stock market intelligence. This paper presents a cloud-native, event-driven pipeline for real-time stock market sentiment analysis and AI-assisted price prediction, designed to operate continuously and at scale. Financial news articles are continuously polled from NewsAPI across ten major technology stocks — AAPL, MSFT, GOOGL, AMZN, META, NVDA, TSLA, NFLX, ADBE, and ORCL — and dispatched to AWS Simple Queue Service (SQS), where AWS Lambda functions apply VADER (Valence Aware Dictionary and sEntiment Reasoner) sentiment scoring. Each article receives a compound sentiment score in the range [?1, 1] and is classified as Positive, Negative, or Neutral. Structured results are then persisted to AWS S3 for downstream consumption. A Flask REST backend integrates Google Gemini AI to generate natural-language analyst summaries by correlating sentiment scores with empirically calibrated stock-specific sensitivity factors. The system exposes three frontend modules built as a lightweight web application named MarketSense: a live Sentiment Dashboard displaying real-time compound scores and sentiment distribution charts, a Scenario Simulator enabling analysts to query AI-generated market impact assessments for hypothetical events, and a News-Driven Price Prediction engine that computes predicted price deltas from live sentiment data. The end-to-end pipeline achieved an average news-to-result latency of 18 seconds and a Flask API response time of 1.4 seconds inclusive of Gemini inference, confirming its suitability for interactive, near-real-time financial decision support. The architecture is fully serverless and horizontally scalable without manual provisioning, making it cost-effective for continuous deployment.

Introduction

Financial markets react quickly to news, but traditional methods based on historical data fail to capture real-time textual signals. To address this, the system uses Natural Language Processing (NLP)—specifically the VADER sentiment model—to classify news headlines as positive, negative, or neutral and estimate market impact within seconds.

The architecture is fully event-driven and serverless, using AWS services (SQS, Lambda, and S3) to handle large-scale news ingestion and processing efficiently. A Flask backend processes sentiment outputs, calculates stock-specific price impact using sensitivity factors, and integrates Google Gemini AI to generate human-like financial summaries. A web dashboard provides live sentiment tracking, scenario simulation, and price prediction visualization.

The system is built for scalability, low latency, and cost efficiency, with automatic scaling during high news activity. It demonstrates how combining rule-based NLP, cloud infrastructure, and generative AI can support real-time financial decision-making.

Conclusion

This work demonstrates a scalable, serverless pipeline that transforms raw news headlines into actionable stock sentiment signals with AI-generated interpretations. The modular architecture enables independent scaling of ingestion, processing, and presentation tiers. Future work will incorporate transformer-based models (e.g., FinBERT [11]) for improved domain-specific sentiment accuracy, integrate real-time WebSocket updates to the dashboard, and add historical back-testing against actual stock returns to validate sensitivity factors quantitatively [14].

References

[1] T. Loughran and B. McDonald, “When is a Liability not a Liability? Textual Analysis, Dictionaries, and 10-Ks,” Journal of Finance, vol. 66, no. 1, pp. 35–65, Feb. 2011. [2] J. Bollen, H. Mao, and X.-J. Zeng, “Twitter mood predicts the stock market,” Journal of Computational Science, vol. 2, no. 1, pp. 1–8, 2011. [3] P. C. Tetlock, “Giving Content to Investor Sentiment: The Role of Media in the Stock Market,” Journal of Finance, vol. 62, no. 3, pp. 1139–1168, 2007. [4] C. J. Hutto and E. Gilbert, “VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text,” in Proc. AAAI Conf. on Web and Social Media (ICWSM), Ann Arbor, MI, 2014. [5] M. Baldini, M. Bagnato, and D. Bruneo, “Serverless Computing in Cloud-Based IoT Systems,” in Proc. IEEE Int. Conf. on Pervasive Computing and Communications Workshops (PerCom), 2017, pp. 577–582. [6] Google DeepMind, “Gemini: A Family of Highly Capable Multimodal Models,” Technical Report, Google, 2023. [Online]. Available: https://deepmind.google/technologies/gemini/ [7] Amazon Web Services, “Amazon SQS Developer Guide.” [Online]. Available: https://docs.aws.amazon.com/AWSSimpleQueueService/latest/SQSDeveloperGuide/ [8] NewsAPI.org, “NewsAPI Documentation.” [Online]. Available: https://newsapi.org/docs [9] S. Mittal and A. Goel, “Stock Prediction Using Twitter Sentiment Analysis,” Stanford University CS229 Project Report, 2012. [Online]. Available: https://cs229.stanford.edu/proj2011/GoelMittal-StockMarketPredictionUsingTwitterSentimentAnalysis.pdf [10] R. P. Schumaker and H. Chen, “Textual Analysis of Stock Market Prediction Using Breaking Financial News: The AZFin Text System,” ACM Trans. Inf. Syst., vol. 27, no. 2, pp. 12:1–12:19, Feb. 2009. [11] D. Araci, “FinBERT: Financial Sentiment Analysis with Pre-trained Language Models,” arXiv preprint arXiv:1908.10063, 2019. [12] Amazon Web Services, “AWS Lambda Developer Guide.” [Online]. Available: https://docs.aws.amazon.com/lambda/latest/dg/welcome.html [13] W. Zhang, C. Ding, and M. Zhang, “Earnings Conference Call Summarisation Using Large Language Models,” in Proc. ACL Workshop on Natural Language Processing in Finance (FinNLP), 2023, pp. 45–54. [14] E. F. Fama, “Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance, vol. 25, no. 2, pp. 383–417, May 1970.

Copyright

Copyright © 2026 Kanchanadevi P, Shivaani Vaithiyanathan Ganesh, HariPriya , Arjun . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.