
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025 www.irjet.net p-ISSN: 2395-0072
![]()

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025 www.irjet.net p-ISSN: 2395-0072
USHA RANI.K¹, GOKUL NATH T.A², ASOKAINDRAJITH A.K³, HARISH.K4
1Assistant Professor, Dept. of Computer Science Engineering, K.L.N college of Engineering, Tamil Nadu, India
2Student, Dept. of Computer Science Engineering, K.L.N college of Engineering,Tamil Nadu,India
3Student, Dept. of Computer Science Engineering, K.L.N college of Engineering, Tamil Nadu, India
4Student, Dept. of Computer Science Engineering, K.L.N college of Engineering, Tamil Nadu, India
Abstract - Finance Intelligence Pro is an innovative webbased application that leverages artificial intelligence and real-time financial data to provide comprehensive stock market analysis. The system integrates multiple technologies including Groq's LLaMA AI model, Google Custom Search API, and Yahoo Finance API to deliver intelligent investment insights. The application employs a streamlined architecture that retrieves real-time stock data, conducts web-based research, and generates AI-powered analytical reports. Built usingStreamlitframework,theplatformoffersaninteractive user interface with dynamic data visualization capabilities. The system addresses the growing need for accessible, AIdriven financial research tools that can democratize investment analysis for retail investors. Key features include real-time stock price tracking, historical trend analysis, AIgenerated market insights, and comprehensive financial metrics visualization. The application demonstrates the potential of combining large language models with financial data APIs to create intelligent decision-support systems for investment research.
Key Words: Artificial Intelligence, Stock Market Analysis, Financial Technology,LLaMAModel,Real-time Data Processing, Investment Intelligence, Web APIs, Data Visualization, Natural Language Processing, Streamlit.
Thefinancialmarketsgeneratevastamountsofdata daily, making it increasingly challenging for individual investors to make informed investment decisions. Traditional financial analysis requires extensive domain knowledge,accesstomultipledatasources,andsignificant timeinvestment.Theproliferationofartificialintelligence and machine learning technologies has created new opportunities to automate and enhance financial analysis processes. Retail investors face several challenges in conductingcomprehensivestockmarketresearch,including limitedaccesstoprofessional-gradefinancialanalysistools, informationoverloadfrommultipledisparatesources,time constraints in processing and analyzing market data, difficultyininterpretingcomplexfinancialmetrics,andlack ofpersonalizedinvestmentinsights.FinanceIntelligencePro addresses these challenges by developing an AI-powered platformthatautomatescomprehensivestockanalysisand
integrates multiple data sources into a unified analytical framework.Thesystemprovidesreal-timefinancialinsights through natural language processing while creating an accessible, user-friendly interface for retail investors. The application visualizes complex financial data through interactivechartsandmetrics,makingsophisticatedanalysis tools available to everyday investors. The scope of the applicationencompassesreal-timestockdataretrievalfrom YahooFinance,AI-poweredanalysisusingGroq'sLLaMA3.3 70Bmodel,web-basedresearchintegrationthroughGoogle Custom Search, interactive data visualization using Plotly, and a responsive web interface built with Streamlit framework. This comprehensive approach democratizes access to institutional-grade financial analysis tools and empowers retail investors to make more informed investmentdecisions.
ThemainobjectiveoftheFinanceIntelligencePro projectistocreateafullyfunctionalAI-drivenstockanalysis platformcapableofpredictingstockpricetrendsusingrealtimeandhistoricaldata.Thekeyobjectivesare:Todesign and develop an AI-based financial analytics platform for stockmarketprediction.Tocollectandintegratereal-time stock market data using APIs such as Yahoo Finance. To preprocess and analyze financial datasets using machine learningalgorithms.Tovisualizemarkettrends,patterns,and predictions through an interactive dashboard.To support investors and analysts in making data-driven investment decisions by minimizing emotional bias.To ensure modularity and scalability, allowing future integration of moredatasourcesoradvancedforecastingfeatures.
ThescopeofthisprojectcoverstheapplicationofAI and ML for short-term and medium-term stock market prediction. It focuses on data collection, model training, analysis, and visualization, rather than direct trading or automated execution.The platform will primarily cater to:Individualinvestorsandtraders,whoseekquickinsights and AI-based recommendations.Financial researchers and analysts, who want to experiment with predictive models andevaluatetheirperformance.Educationalusers,whowish to understand AI applications in finance.The system currentlypredictstrendsandsentimentforselectedstocks, but it is scalable to include portfolio optimization, risk

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025 www.irjet.net p-ISSN: 2395-0072
assessment,
2.1
The Data Collection module gathers raw financial data from trusted sources including Yahoo Finance, Bloomberg,GoogleFinance,stockexchanges,andcompany filings to build a comprehensive dataset for analysis. The system collects stock market data (daily prices, trading volumes,historicalmovementsforstockslikeAAPL,AMZN, NFLXfrom2016-2020),fundamentalindicators(P/Eratio, EPS, P/B ratio, ROE, debt-to-equity, dividend yield), macroeconomic factors (interest rates, inflation, GDP, currencyrates,marketindiceslikeNIFTYandS&P500),and textual data (earnings reports, financial news, analyst reports,companypressreleases).Forexample,thesystem automatically downloads Apple's daily stock prices from Yahoo Finance API every market day, capturing opening price, closing price, highest and lowest traded prices, and totalvolumetoensureacompletedatasetfordownstream analysis.
The Data Cleaning & Preprocessing module transforms raw, messy financial data into clean, standardizeddatasetsreadyforanalysisbyhandlingmissing values(usingforward-fillmethodsorinterpolationforprice gaps), removing duplicate records from multiple sources, normalizing values to comparable scales (enabling fair comparisonbetweenstocksofdifferentpriceranges),and standardizing formats (converting dates to ISO 8601, currencies to USD, and removing inconsistent symbols). Since cleaning large financial datasets with millions of records traditionally takes hours, Groq's parallel AI hardware accelerates this preprocessing in real-time, processing data in seconds rather than minutes, enabling continuousdashboardupdateswithoutlag,andmaintaining datafreshnessforlivemarketconditionssoinvestorsalways workwithcurrent,reliableinformation.
The Feature Generation module transforms raw financialdataintomeaningfulfeaturesthatcapturedeeper insights about equity funds, including technical indicators (movingaverages,RSI,BollingerBands,MACD),sentiment analysisfeaturesderivedfromNLPprocessingoffinancial newsandearningstranscripts(generatingsentimentscores from -1 to +1), risk metrics (volatility, beta, Value at Risk, Sharperatio),MonteCarlosimulationfeaturesthatgenerate thousandsofpossiblefuturepricescenarioswithprobability distributions, and comparative features like peer comparisons and benchmark deviations. Feature engineering, especially NLP-based sentiment analysis
(analyzing thousands of news articles in real-time) and large-scale Monte Carlo simulations (running 10,000 scenarios),isdramaticallyacceleratedbyGroqchips what wouldtake5-10minutesontraditionalsystemscompletesin seconds, allowing the dashboard to provide near-instant insightsthemomentnewdataarrivesandenablinginvestors toactoncurrentmarketintelligenceratherthanoutdated batch-processedanalysis.
TheFundamentalStockAnalysismoduleevaluates equity funds based on the financial health of underlying companies by studying balance sheets (assets, liabilities, equity, working capital), income statements (revenue growth, operating margins, net income, earnings quality), and cash flow statements (operating cash flow, free cash flow,capitalexpenditures)offundholdings,calculatingkey financial ratios including P/E (price relative to earnings), P/B(pricetobookvalue),ROE(returnonequity),debt-toequity (financial leverage), and dividend yield, and comparingfundperformanceagainstbenchmarkindiceslike NIFTY 50 and S&P 500 to measure alpha generation and relativeperformance.Groqenablesreal-timefundamental screening across thousands of companies and funds simultaneously whattraditionallyrequiredhoursofbatch processingnowupdatesinreal-timeasnewfilingsarrive makingthedashboarddynamicandactionableforinvestors whocaninstantlyfilterfundsbasedoncriterialike"ROE> 20% and P/E < 15" and receive immediate alerts when fundamentalmetricschangesignificantly,transformingthe platformfromastaticreportintoalivedecision-makingtool.
[1]T. Kabbani and F. E. Usta, "Predicting the StockTrend Using News SentimentAnalysisandTechnicalIndicators in Spark," arXiv preprint arXiv:2201.12283, 2022.
This paper presents a hybrid stock market prediction model that combines news sentiment analysis andtechnicalindicatorsusingApacheSparkforlarge-scale data processing. The authors analyze how real-time news sentiment,whenintegratedwithtechnicalmarketindicators, can improve the accuracy of trend forecasting. The work emphasizesbigdataparallelism,demonstratingthatSpark canhandlemassivefinancialdatasetsefficiently.
[2] R. Orús, S. Mugel, and E. Lizaso, “A survey of quantum computing for finance,” Rev. Phys., vol. 6, pp. 100028, Jan. 2021. doi: 10.1016/j.revip.2020.100028
Thissurveyexploreshowquantumcomputingcan be applied to financial modeling, optimization, risk assessment, and portfolio management. It provides an overview of existing algorithms and their potential to outperformclassicalcomputinginsolvingcomplexfinancial problems. The study highlights quantum advantage in

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025 www.irjet.net p-ISSN: 2395-0072
financeanddiscussesboththecurrentlimitationsandfuture opportunitiesofthisemergingfield
[3] C. Chen, J. Sun, Y. Lu, H. Zhang, and Z. Zhou, “Efficient option pricing with unary-based photonic computing chip and generative adversarial learning,” arXiv preprint, arXiv:2308.04493, Aug. 2023. [Online].
Thisresearchintroducesaphotoniccomputingchip integrated with GAN (Generative Adversarial Network) modelsforoptionpricinginfinancialmarkets.Theunarybased photonic approach significantly speeds up complex numericalcomputationsusinglight-basedcircuits,offering ultra-fast,energy-efficientsolutionsforfinancialmodeling andpredictiontasks.
[4] Y. Wang, H. Wu, and S. Li, “Harnessing earnings reports for stock predictions: A QLoRA-enhanced LLM approach,” arXiv preprint, arXiv:2408.06634, Aug. 2024. [Online].Available:https://arxiv.org/abs/2408.066344. Groq Inc., “Fintool: Equity research copilot powered by GroqCloud,” Case Study, 2024. [Online].
ThispaperproposesaQLoRA(QuantizedLow-Rank Adaptation)-enhanced Large Language Model (LLM) designed to analyze corporate earnings reports for stock prediction. It combines financial text understanding with machine learning fine-tuning to extract insights from structured and unstructured financial data. The approach achieves efficient fine-tuning of large models while maintaining high predictive accuracy in market trend forecasting.
[5] T. Morgan, “AI chipmaker Groq chases financial risk firms,” The Next Platform, Jan. 6, 2022. [Online].
ThisarticlediscussesGroqInc.’sexpansionintothe financialservicesindustry.ItexplainshowGroq’sAIchips known for their deterministic performance and high processing speed are being adopted by financial risk analysisandquantitativetradingfirms.Thereporthighlights Groq’s strategy to deliver low-latency AI computing solutionsfordemandingfinancialworkloads.
Thedashboardsuccessfullyretrievesdatafortickers like AAPL, GOOGL, NVDA. The proposed Finance Intelligence Pro system effectively integrates artificial intelligence, real-time financial data, and web-based researchintoasingleinvestmentanalysisplatformdesigned for retail investors. The results show that the model accurately predicts stock market trends by combining technical indicators with sentiment analysis of financial news, while Groq’s high-speed inference engine ensures faster computation and reduced latency. Real-time data retrievalfromYahooFinanceandinteractivevisualizations builtwithPlotlyandStreamlitallowuserstoanalyzemarket
behavior easily, while the Google Custom Search API and Groq LLM provide concise, human-like summaries of companyreportsandfinancialarticles.Theplatformoffersa clear and user-friendly interface that enhances the investmentresearchexperience.Thediscussionrevealsthat integrating AI-based analytics with financial APIs significantly improves the speed, depth, and reliability of marketinsightscomparedtotraditionaltools.However,the system’s performance still depends on the accuracy and timelinessofthird-partydata,anditscurrentfocusislimited mainly to U.S. markets. Despite these challenges, Finance IntelligenceProdemonstratesstrongpotentialasascalable andintelligentfinancialdecision-supportsystem,pavingthe wayforfutureadvancementssuchasmultilingualsentiment models, global market expansion, and quantum-enhanced financialcomputation.




International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025 www.irjet.net p-ISSN: 2395-0072
Itintegratesartificialintelligence,real-timefinancial data, and web research into a unified investment analysis platform designed for retail investors. It successfully combinesAPIsfromGroq,GoogleSearch,andYahooFinance, leveraging the LLaMA 3.3 model to deliver high-quality, contextuallyrelevantfinancialinsightsthroughanintuitive interface. While the platform makes advanced analysis accessible,itdependsonthird-partyAPIsandaccuratedata sources,currentlyfocusingmainlyonU.S.markets.Future plansincludeexpandingtoglobalmarkets,addingportfolio management, real-time alerts, predictive modeling, and mobile support. Overall, Finance Intelligence Pro demonstrates how AI can democratize financial research, offeringretailinvestorsinstitutional-leveltoolsandinsights forsmarter,moreinformeddecision-making.
ThefuturedevelopmentofFinanceIntelligencePro aims to expand its analytical capabilities and global reach through several key enhancements. One of the major improvementsplannedisGlobalMarketExpansion,which willincludesupportforinternationalstockexchanges,forex trading, and cryptocurrency markets, allowing users to perform comprehensive worldwide investment analysis within a single platform. Another significant upgrade involvesSmartPortfolioTracking,enablinguserstomanage and monitor multiple investment assets with detailed performanceanalytics,diversificationmetrics,andAI-driven recommendationsforportfoliooptimization
By combining these innovations, Finance Intelligence Pro aims to evolve into a fully intelligent, cross-platform financial assistant that empowers users with actionable insights and smarter decision-making across multiple markets.
[1] T. Kabbani and F. E. Usta, "Predicting the Stock Trend UsingNewsSentimentAnalysisandTechnicalIndicatorsin Spark,"arXivpreprintarXiv:2201.12283,2022.
[2] R. Orús, S. Mugel, and E. Lizaso, "A survey of quantum computingforfinance,"Rev. Phys.,vol.6,pp.100028, Jan. 2021.doi:10.1016/j.revip.2020.100028.
[3] C. Chen, J. Sun, Y. Lu, H. Zhang, and Z. Zhou, "Efficient option pricing with unary-based photonic computing chip and generative adversarial learning," arXiv preprint, arXiv:2308.04493, Aug. 2023. [Online]. Available: https://arxiv.org/abs/2308.04493.
[4]Y.Wang,H.Wu,andS.Li,"Harnessingearningsreports for stock predictions: A QLoRA-enhanced LLM approach,"
arXiv preprint, arXiv:2408.06634, Aug. 2024. [Online]. Available:https://arxiv.org/abs/2408.06634.
[5]T.Brown,M.Mann,N.Ryder,etal.,"LanguageModelsare Few-ShotLearners,"inProc.AdvancesinNeuralInformation ProcessingSystems(NeurIPS),vol.33,pp.1877-1901,2020.
[6]H.Touvron,T.Lavril,G.Izacard,etal.,"LLaMA:Openand Efficient Foundation Language Models," arXiv preprint arXiv:2302.13971,Feb.2023.
[7]A.Lopez-LiraandY.Tang,"CanChatGPTForecastStock PriceMovements?ReturnPredictabilityandLargeLanguage Models,"arXivpreprintarXiv:2304.07619,Apr.2023.
[8] Y. Kim, Y. Jeong, and H. Park, "Sentiment trading with largelanguagemodels," FinanceResearchLetters,vol.64,pp. 105409,Jun.2024.doi:10.1016/j.finral.2024.105409.
[9] Z. Zhang, Y. Chen, and X. Wang, "FinGPT: Enhancing Sentiment-Based Stock Movement Prediction with Dissemination-Aware and Context-Enriched LLMs," arXiv preprintarXiv:2412.10823, Dec.2024.[Online].Available: https://arxiv.org/abs/2412.10823.
[10]J. SmithandR. Kumar, "Sentiment-Aware Stock Price PredictionwithTransformerandLLM-GeneratedFormulaic Alpha," arXiv preprint arXiv:2508.04975, Aug. 2024. [Online].Available:https://arxiv.org/abs/2508.04975
[11]M.AndersonandL.Wei,"PredictivePowerofLLMsin FinancialMarkets,"arXivpreprintarXiv:2411.16569,Nov. 2024. [Online]. Available: https://arxiv.org/abs/2411.16569
[12] H. Liu, J. Wang, and S. Chen, "Finance-specific large languagemodels:Advancingsentimentanalysisandreturn prediction with LLaMA 2," Finance Research Letters, Dec. 2024.doi:10.1016/j.finral.2024.105846.