Graphical Summary of Annual Reports

[dayton-outar]

Take material from the annual reports to build a graphical presentation for each company and put them into a slide show that can be accessed when drilling into details of stocks.

• [ ] Study different types of annual reports to identify patterns of information between them
• [ ] List the possibilities for automation. Outline all the steps. Both manual and automated.
• [ ] Parse PDF of Annual Report
  • [ ] Use NLP or LLM to read narratives and summarize content
    • OpenAI API
    • Langchain
    • Kor
  • [ ] Extract images using automation

[dayton-outar]

Investors typically look for several key pieces of information when reading annual reports of a stock exchange. These include:

1. Financial Performance: Investors closely examine financial statements such as the income statement, balance sheet, and cash flow statement to assess the company's financial health, revenue growth, profitability, and cash flow generation.

2. Management Discussion and Analysis (MD&A): This section provides management's perspective on the company's financial results, operational performance, risks, and future prospects. Investors pay attention to management's insights and outlook.

3. Business Strategy: Investors seek to understand the company's business model, competitive positioning, market trends, and growth strategies outlined in the annual report. This helps them assess the company's long-term viability and growth potential.

4. Risk Factors: Annual reports typically disclose various risks that could impact the company's operations and financial performance. Investors evaluate these risks to gauge the potential downside and assess the company's risk management practices.

5. Corporate Governance: Investors look for information on the company's corporate governance practices, including board composition, executive compensation, and any related-party transactions. Strong corporate governance signals management's commitment to transparency and accountability.

6. Auditor's Report: Investors review the auditor's report to assess the reliability of the financial statements and the effectiveness of the company's internal controls.

7. Footnotes and Disclosures: Annual reports contain footnotes and disclosures that provide additional context and detail on various aspects of the company's operations, accounting policies, and significant transactions. Investors pay attention to these notes to gain a deeper understanding of the company's financials.

8. Financial Ratios and Metrics: Investors analyze key financial ratios and performance metrics (e.g., earnings per share, return on equity, debt-to-equity ratio) to benchmark the company's performance against its peers and industry standards.

By scrutinizing these elements in annual reports, investors can make informed decisions about whether to buy, hold, or sell a company's stock.

[dayton-outar]

Checklist for IPO Investment:

• The text outlines a five-point checklist for analyzing IPOs:
  • Objects of the Offer: Explains how the company intends to utilize the funds raised.
  • About the Company: Provides insights into the company's background, financials, and industry overview.
  • Management: Evaluates the management team's experience and qualifications.
  • Legal and Other Information: Examines any legal issues or liabilities the company may have.
  • Valuation: Offers a simplified method for evaluating the company's valuation, including comparing the PE ratio with industry peers.
  • Understanding the Risks: Identifies potential risks associated with the company, market, and industry.

[dayton-outar]

While the Auditor's Report, Financial Performance and ratios can be had from the part of the annual report dedicated to financial statements, the initial part of the document usually contains the narratives about the company and the management's expectations of future cashflows and market behaviour.

Narratives about the company from the executives and board members need to be extracted to put in the context of,

1. Management Insights from Management Discussions and Analysis
2. Business Strategy, which includes business model used and market research discovered by the management team
3. Risk Factors, which involves the risks known by the management team
4. Corporate Governance which is the politics of the institution. The persons involved, the transactions they conduct, the skills that they bring to the operations.

[dayton-outar]

Important details concerning the company that requires extraction includes,

1. Company Profile. A summary that describes the company and its operations.
2. Company's Mindset
   1. Mission (usually stated in a Mission Statement)
   2. Vision (if provided)
   3. Purpose (if provided)
   4. Values (if provided)
   5. Branding (if provided)
   6. Policies (included in Policy Statements)
3. Report on company's environment and social contribution (ESG). Corporate Social Responsibility.
4. Report from board member, usually the Chairman
5. Report from a member of the executive, usually the CEO
6. Report on Corporate Governance changes
7. List of directors and their profiles
8. Directors' Report
9. List of executive (or management team) and their profiles
10. Board attendance report
11. Disclosure of shareholdings
    1. Top Ten Largest Shareholders
    2. Shareholdings of Directors and Connected Party
12. Shareholder's Report or Management Discussion
13. Risk Management (if provided)

Note well. Carreras 2022 Annual Report was very extensive covering marketing and human resource details. Is that information worthwhile to investors? Maybe TMI.

[dayton-outar]

Some things to consider are awards and recognition. NCB had a section for this in their Corporate Governance report.

[dayton-outar]

Since this requires a graphical treat for the investors, as it is intended by the publishers of this reports. A number of photos are found in the document. Items that must be extracted includes,

1. Full page photos
2. Profile portrait photos of board members and executives (management team)
3. Cover graphic for the annual report

The color scheme could be noted for use in the presentation of the summarized version on the app.

[dayton-outar]

Note well. EduFocal (LEARN) 2022 Annual Report is a fully graphic PDF and will require OCR to extract textual information.

Also. Proven Jamaica (PROVENJA) 2019 Annual Report is a 2-column document that may present issues to the OCR.

[dayton-outar]

Do we need to extract information on corporate structure? Or can this be done manually? (It may be easier to do this manually as how the company profile is done in Shillings now) ... Ah! Corporate Structure, comes under Company Profile.

[dayton-outar]

So, we know the snippets of information and graphics that we need to extract from the narrative of an annual report but the challenge now comes with the most optimal way to extract the information. To further explain this challenge, the PDF documents containing the narratives can sometimes included images rather than a mix of image and text. In such a case where the narratives are in an embedded image, an OCR will be required to extract the text. Not all annual reports are the same. Some annual reports are prepared in such a manner that the text can be easily extracted with a pdftotext utility tool, while some reports have made all pages into an embedded image. Some annual reports have a mix of both.

A pseudocode describing a process to extract information from multiple composition is required at this stage. First priority is to find a tool that can query a PDF document in a manner that SQL can query a database. Can the information provided in this tutorial video, Learn How To Query Pdf using Langchain Open AI in 5 min, help.

The ChatPDF online application serves as an example of the feature needed to embellish the graphical display of the annual report. So, with prompt engineering the line items in the financial statements can be classified and prepared for entry into the shillings database.

This project will indeed operate in tandem with issue #4 and thus shares interest in a an extraction process that can intelligently classify the information provided in the annual reports that come in many variations.

[dayton-outar]

Here are some resources that has been helping in the research,

• Repo: Langchain Tutorials; Playlist: LangChain How to and guides
• Repos: Langchain Examples; Playlist: LangChain for Gen AI and LLMs
• Using langchain for Question Answering on Own Data

[dayton-outar]

Expert.js Repo is an LLM tool worth looking into.

[dayton-outar]

What needs to be tested is the extraction of text on annual reports like 2022 Annual Report of Edufocal that has significant number of embedded images containing text.