SmartFinance AI: Intelligent NASDAQ-100 Analytics Platform
The Nasdaq-100 is a stock market index that tracks the equity shares of 100 of the largest non-financial firms, both domestic and foreign, that trade on the Nasdaq exchange. It uses a modified capitalization-weighted structure. The average year-over-year return of the Nasdaq-100 over the last 10 years is 20.7% with a cumulative return of 558.8%. The compound annual growth rate (CAGR) is 16.93% with a standard deviation of 17.69%, reflecting some volatility and market risk compared to the S&P 500, which has a CAGR of 10.98% with a standard deviation of 15.24%. The recent struggles of hedge funds, which saw investors pull over $150 billion in five years and short sellers lose nearly $195 billion in 2023. However, as of this writing, the Nasdaq-100 has reached an all-time high in 2024, with a year-to-date increase of over 35%. Such erratic swings highlight the critical need for better financial tools for investors. A well-designed financial dashboard application, integrated with Large Language Models (LLMs), can help hedge funds and investors make more informed decisions, avoid costly misjudgments, and navigate volatile markets more effectively. By leveraging LLMs for real-time document analysis, market sentiment tracking, and automated risk assessment, investors can process vast amounts of financial data more efficiently while uncovering deeper insights that might be missed through traditional analysis alone.
Data Overview
We obtained our dataset from Kaggle, which contains Nasdaq-100 data from 2017 to 2023, and their associated financial indicators. These indicators are essential for evaluating the financial health and performance of companies, helping investors make informed decisions. The 31 financial metrics included in the dataset are listed here:
- Asset Turnover
- Buyback Yield
- CAPEX to Revenue
- Cash Ratio
- Cash to Debt
- COGS to Revenue
- Beneish M-Score
- Altman Z-Score
- Current Ratio
- Days Inventory
- Debt to Equity
- Debt to Assets
- Debt to EBITDA
- Debt to Revenue
- E10 (by Prof. Robert Shiller)
- Effective Interest Rate
- Equity to Assets
- Enterprise Value to EBIT
- Enterprise Value to EBITDA
- Enterprise Value to Revenue
- Financial Distress
- Financial Strength
- Joel Greenblatt Earnings Yield (by Joel Greenblatt)
- Free Float Percentage
- Piotroski F-Score
- Goodwill to Assets
- Gross Profit to Assets
- Interest Coverage
- Inventory Turnover
- Inventory to Revenue
- Liabilities to Assets
Key Financial Metrics Explained
These are essential financial metrics used in the analysis of Nasdaq-100 companies, along with their definitions, formulas, and typical ranges for good and bad values.
1. Price to Earnings (P/E) Ratio
The P/E ratio measures how much investors are willing to pay per dollar of earnings. It is defined as P/E Ratio = Price per Share / Earnings per Share (EPS).
- A good P/E ratio is typically between 15 and 25. It suggests a company is fairly valued.
- A low P/E ratio (< 10) may indicate the company is undervalued, or it could suggest future earnings trouble.
- A high P/E ratio (> 30) may indicate a company is overvalued or has high future growth expectations.
Investors should compare the P/E ratio with those of other companies within the same sector, as well as across different sectors and the broader market, to determine whether the P/E ratio is favorable or not.
2. Debt to Equity Ratio
The Debt to Equity ratio measures a company’s financial leverage and compares its total debt to shareholders' equity. It is defined as Debt to Equity Ratio = Total Debt / Shareholders' Equity.
- A good debt to equity ratio is typically between 0.5 and 1.0, indicating balanced financial leverage.
- A higher ratio (> 2.0) suggests the company is heavily leveraged and may face higher financial risk.
- A low ratio (< 0.5) may indicate under-utilization of debt for financing growth.
A “good” debt-to-equity (D/E) ratio varies by industry, but generally, a ratio below 1 is considered safe, while a ratio of 2 or higher is seen as risky, with some industries like utilities, consumer staples, and banking typically having higher D/E ratios.
3. Revenue Growth YoY
Revenue Growth YoY shows the percentage increase or decrease in a company’s revenue compared to the previous year. It is defined as Revenue Growth YoY = ((Current Year Revenue - Previous Year Revenue) / Previous Year Revenue) * 100%.
- A good revenue growth rate is generally 10-20% YoY, indicating strong business expansion.
- A negative or low growth rate can signal financial challenges or stagnant growth.
4. Beneish M-Score
The Beneish M-Score is a model that detects potential earnings manipulation by analyzing various financial ratios. A company’s likelihood of manipulation is defined by the formula M-Score = Complex formula involving multiple financial ratios.
- A score below -1.78 is safe and indicates little to no manipulation risk.
- A score above -1.78 may suggest potential earnings manipulation.
5. Altman Z-Score
The Altman Z-Score predicts a company’s bankruptcy risk based on various financial metrics. It is calculated using the formula Z-Score = 1.2A + 1.4B + 3.3C + 0.6D + 1.0E, where A to E represent various financial ratios.
- A good Z-Score is above 3.0, indicating low bankruptcy risk.
- A score between 1.8 and 3.0 suggests moderate risk.
- A score below 1.8 signals high bankruptcy risk.
6. Asset Turnover
Asset Turnover measures how efficiently a company uses its assets to generate sales. It is defined as Asset Turnover = Revenue / Total Assets.
- A higher asset turnover (> 1.0) indicates effective use of assets to drive revenue.
- A low asset turnover (< 0.5) suggests inefficient asset utilization.
7. Gross Profit to Assets
Gross Profit to Assets assesses how effectively a company generates profit from its total assets. It is defined as Gross Profit to Assets = Gross Profit / Total Assets.
- A good ratio (> 20%) indicates strong profit generation relative to assets.
- A low ratio (< 10%) signals inefficient asset profitability.
8. Cash Ratio
The Cash Ratio measures a company’s ability to pay off short-term liabilities with its cash and cash equivalents. It is defined as Cash Ratio = Cash and Cash Equivalents / Current Liabilities.
- A cash ratio above 1.0 is favorable, indicating the company can easily cover short-term liabilities.
- A low cash ratio (< 0.5) suggests potential liquidity concerns.
9. Current Ratio
The Current Ratio indicates a company’s ability to meet its short-term liabilities with its short-term assets. It is defined as Current Ratio = Current Assets / Current Liabilities.
- A current ratio between 1.5 and 3.0 is considered healthy, indicating sufficient liquidity.
- A ratio below 1.0 may signal liquidity issues.
10. Debt to Assets Ratio
The Debt to Assets ratio shows the proportion of a company’s assets that are financed through debt. It is defined as Debt to Assets Ratio = Total Debt / Total Assets.
- A good ratio is below 50%, indicating the company is not overly reliant on debt.
- A high ratio (> 60%) suggests the company is heavily dependent on debt, which can be risky.
Streamlit: Seamless Dashboard Deployment Integrated with LLM
Streamlit is an innovative framework that converts data scripts into shareable web apps in minutes using only Python. This means we don't need any prior front-end development skills to create dynamic and visually beautiful applications. Since Snowflake acquired Streamlit in 2022, the company's functionality has been integrated into Snowflake's enterprise-grade data solutions, which are used by some of the world's largest corporations. Streamlit's straightforward API enables us to quickly design and deploy web apps, making it ideal for data scientists and analysts who are more familiar with Python than front-end technologies. Furthermore, Streamlit provides real-time data visualization, which is essential for developing dynamic dashboards that update as new data arrives. This function is especially useful for measuring financial metrics, tracking performance indicators, and other time-sensitive applications.
Streamlit is ideal for generative AI applications because of its capacity to handle real-time data and interactive components. That’s particularly useful now that Large Language Models (LLMs) gain popularity. The forthcoming release of chat components will expand its features, enabling us to build sophisticated LLM-powered apps that make it simple to integrate Streamlit with the burgeoning ecosystem of new technologies. For example, we may use LangChain's callback system to obtain insights into an LLM's thought process and then add intermediate step information to your Streamlit app with a single command.
In conclusion, Streamlit is one of the greatest tools for creating dashboards and LLM apps due to its ease of use, real-time visualization capabilities, interactive features, and easy connection with the ecosystem. Whether we are creating a financial dashboard or a sophisticated AI-powered app, Streamlit provides the tools we need to succeed.
LLM Enhanced Market Analysis
Our platform leverages two powerful LLM integrations to deliver comprehensive financial insights. We've implemented GPT-4 as a financial expert system, enabling users to ask sophisticated questions about market trends, financial metrics, and investment strategies, receiving detailed, context-aware responses backed by current market knowledge. Additionally, we've integrated Langchain's Conversational Retrieval Chain for advanced document processing, allowing users to upload multiple financial documents (such as annual reports, SEC filings, and market analyses) and extract relevant insights through natural language queries.
This dual approach combines GPT's analytical capabilities with Langchain's document processing power, enabling users to not only analyze current market data but also process vast amounts of historical and company-specific documentation efficiently as shown in Figure 1. The conversational nature of both systems makes complex financial analysis accessible and intuitive, whether you're querying live market data or analyzing hundreds of pages of financial documents. Our platform offers flexible model selection – from GPT-3.5 for rapid analysis to o1 for complex reasoning, DALL-E for visualization, and TTS for audio generation – allowing users to choose the most suitable model for their specific analytical needs.
NASDAQ-100 Financial Dashboard: Real-Time Insights and Analysis
The figures below are interactive, allowing you to hover over them to view detailed labels and values. The sidebar and main panel architect of the app is shown below in Figure 2.
The sidebar features an expandable user guide, dropdown menus for sector, subsector, and companies, and sliders for selecting years. The filtered data is displayed in a table below, which users can download for further analysis. By default, it shows financial indicators for all companies over the entire date range, but users can refine their selection using the dropdown menus or sliders.
The main panel highlights top companies based on the highest Price-to-Earnings (P/E) ratio, highest year-over-year revenue growth, and lowest Debt to Equity (D/E) ratio for the most recent year in selected years. Long-term metrics such as the M-score, Z-score, and CAGR averaged over the past five years are displayed in donut plots. Hovering over the brighter sections of these plots reveals the percentage of companies to avoid based on their M and Z-scores and five-year CAGR.
For more details on the metrics and company names, expand the “Read More about Investment Avoidance and Average Return on Investment” section.
Figure 3 illustrates trends in performance and profitability metrics, including asset turnover, gross profit to assets, year-over-year revenue, and EPS growth for selected companies over a specified time period.
Figure 4 illustrates trends in liquidity and cash management metrics, including the cash ratio and current ratio for selected companies over a specified time period.
Figure 5 depicts debt and leverage metrics, highlighting trends in debt-to-equity and debt-to-assets ratios over a selected time period.
Figure 6 illustrates investment risk by displaying the average Beneish M-Score and Altman Z-Score for a given company over a selected time period. Companies with an average M-Score above -1.78 are highly likely to be manipulators, while those with a Z-Score below 1.81 are at risk of heading towards bankruptcy.
Conclusion
By leveraging a cloud-hosted Streamlit application integrated with advanced LLM capabilities, we've created a sophisticated financial analysis platform that excels at processing both market data and historical financial information. The app's interactive dashboards, accessible across all devices, combine traditional financial metrics with AI-powered insights, enabling users to analyze market trends, company performance, and risk indicators through an intuitive interface. Our integration of GPT for financial analysis and Langchain for document processing allows investors to not only track conventional metrics but also gain AI-enhanced insights from unstructured data sources such as financial reports, news feeds, and market commentary. This comprehensive approach, combining real-time data processing with advanced analytics, empowers investors to make data-driven decisions across various investment horizons – from intraday trading opportunities to long-term value investments – while maintaining a clear understanding of risk-reward dynamics in volatile market conditions.
Future Work
Our current dataset is static and cut off at 2023. However, by integrating real-time data such as NASDAQ-100 total market cap, money inflow and outflow volumes, stock prices, and general economic indicators, we can provide real-time updates. This integration allows us to track short-term market projections and long-term trends, offering insights into the direction companies are heading. Implementing RAG architecture in our system enables sophisticated market analysis by utilizing a vector store of historical market events and their corresponding price movements. When new market-moving news emerges, our RAG system can instantly retrieve similar historical scenarios from its knowledge base, analyze the market's previous reactions, and generate predictions about potential price movements. This system becomes particularly powerful when combined with real-time sentiment analysis of financial news, social media feeds, and market commentary, allowing for rapid identification of potential market-moving events and their likely impact on specific sectors or companies within the NASDAQ-100.
Acknowledgement
I would like to express my gratitude to Vivian S. Zhang for introducing me to Streamlit for building dashboards and interactive applications powered by LLMs. I also want to thank Vinod Chugani for suggesting the fascinating NASDAQ-100 dataset and encouraging me to explore new tools in the pursuit of leveraging LLMs in applications.
If you enjoyed reading my blogpost, please follow and connect me on LinkedIn for collaboration, networking, and more insightful content.
Click here to view the app.
Click here to view the code in GitHub Repository.
About Author
