SIA

Interactive Stock Investment Advisor (SIA) that provides users with company/stock information through a Q&A format.

Demo video

Project Proposal: Stock Investment Advisor (SIA)

Getting started

1. First, install depedent libraries

$ pip install -r requirements.txt

2. Prepare tokens for Discord bot and Vertex AI (Google) in your shell script!

$ vi ~/.zshrc
export DISCORD_TOKEN="Paste your bot token!"
export VERTEX_AI_API_KEY="Paste your Vertex AI API key!"

3. Run chatbot!

$ python chatbot.py

Goal

To create an interactive stock investment advisor that provides users with company information through a Q&A format. Our goal is to empower investors and financial enthusiasts with accurate, timely, and insightful information.

Features

• Access to Financial Statements/Reports of Tech Companies
  -> Get instant access to the latest financial reports and statements from tech companies.

• Key Points from Financial Statements/Reports
  -> Stay up-to-date on the key performance indicators (KPIs) that matter most for your investments.

• Summarization of the Latest Related News with Sentiment Analysis on the News
  -> Receive expertly curated news summaries, complete with sentiment analysis to help you make informed investment decisions.

• Organized Report Detailing Company Performance
  -> Get a comprehensive overview of each company's performance, including metrics and insights that matter most for your investments.

• Insights on Whether the Company’s Stock is Growing or Declining
  -> Make data-driven decisions with our expertly crafted stock performance insights.

Scope

Our project will focus on the following key elements:

• Coverage: We will cover tech companies in the top 50, providing a comprehensive overview of their financial health and market performance.

• Time Period for Report Summarization: We will summarize reports for the current year 2024.

• Time Period for News Summarization: We will provide news summaries up to the current date, with a focus on the last 12 months.

• APIs and News Sources: Our API integration will be limited to no more than two sources each, ensuring efficient data retrieval and minimal latency.

Team

Meet our talented team of engineers and researchers:

Chaeeon Lim

• News Information Extraction
  • Data Collection: Parsing financial reports, SEC filings, and earnings call transcripts to collect valuable data.
  • Event Extraction: Identifying key metrics, events, and insights from text data.
  • Named Entity Recognition: Extracting named entities like company names, people, locations, etc.
• Document Classification
  • Topic-specific Feature Engineering: Sorting news articles by topic (e.g. mergers, earnings, leadership changes)
  • Text categorization: Classifying company filings by type
  • Relevance Scoring: Identifying relevant vs. irrelevant information.
• Dialogue Assistant
  • Financial summary Generation: Summarizing key financial metrics and performance
  • Recommendation Generation: Generating portfolio reviews and market updates
  • Natural Language Generation: Producing customized client communications

Uddesh Santosh Kumar Singh

• Report Information Extraction
  • Named Entity Recognition: Identifying the various metrics and entities mentioned in Financial Reports.
  • Information Retrieval: Find and retrieve structured information from long reports.
  • Data Mining: Locate key words and metrics in unstructured data within reports.
• Sentiment Analysis
  • Sentiment Classification: Classify text into predefined sentiment categories, such as positive, negative, or neutral
  • Metric-Based Sentiment Understanding: Understand the emotions conveyed by the structured data. Example- Missed Target/ On Target/ Exceeded Target for net profits
• Dialogue Assistant
  • Intent Recognition: Classify the user's intent, such as retrieving financial data, asking for stock performance, or requesting sentiment analysis of a news event
  • Query Expansion and Paraphrasing: Rephrase user queries to improve the accuracy of results, and refine responses based on user input
  • Response Generation: Generate coherent and informative responses to user queries, using data from financial reports, sentiment analysis, and other sources

Data Sources

Our project will leverage the following open-source APIs and news sources:

• Available Open-Source APIs for financial news
• Investing.com/Companies investor relations website

Open-Source APIs

• Sentiment Analysis, public APIs
• News APIs: Associated Press, New York Times, Bloomberg API
• Financial APIs: News api, Mediastack, Marketaux