Building a Stock Scoring Engine Without Writing Code

What is Gruntech Invest?

Gruntech Invest is an automated stock analysis platform that scores over 7,000 US stocks using a systematic, data-driven approach. Instead of relying on gut feelings or tips from friends, this system analyzes financial fundamentals, calculates an objective score (0-100), and uses AI to explain the results in plain English.

The goal: Help everyday investors (like my classmates or parents) understand which stocks might be worth researching further — without needing a finance degree.

The Challenge

Most stock analysis tools are either too complicated (Bloomberg terminals cost $20,000/year) or too simplistic (just showing price charts). I wanted to build something in between: powerful enough to be useful, but simple enough that anyone could understand it.

The Solution

A simple two-layer scoring system:

1. Classic Investor Factors: All 7,000+ stocks are scored using 6 proven investment factors (quality, health, value, growth, sentiment, momentum) — the same factors used by professional investors.
2. AI Scoring (Top 20): The highest-scoring 20 stocks get a full AI analysis from Claude. The AI provides:
   • Its own score (0-100) — an independent AI assessment
   • Conviction level — how confident the AI is
   • Buy/Sell signal — Strong Buy, Buy, Hold, Sell, Strong Sell
   • Bull & Bear cases — best and worst scenarios
   • Plain-English explanation — for beginners
   • Earnings surprise analysis — warns if the company missed expectations
   • Cyclicality warnings — flags if profits might be at peak cycle and unsustainable

The final "Combined Score" blends classic analysis (70%) with AI judgment (30%) — getting the best of both: data-driven precision plus AI's ability to see the bigger picture.

🤖 Zero Code Written — Just AI "Vibe Coding"

This entire project was built using Cursor and Kiro — AI-powered coding tools that let you build software by having a conversation. I described what I wanted, and the AI wrote all the code.

What is "Vibe Coding"?

Instead of learning programming languages and syntax, I had conversations with AI tools:

• "Create a system that fetches stock data and scores them based on quality factors"
• "The page is loading too slowly, can you optimize it?"
• "Add AI explanations for the top stocks so beginners can understand them"
• "Deploy this to AWS and set up automatic weekly updates"

The AI understood my intent and generated the code, fixed bugs, and even deployed everything to the cloud.

Tools I Used:

☁️ Deploying to AWS (Real-World Cloud)

This isn't just a school project running on my laptop — it's a production system deployed on Amazon Web Services (AWS), the same infrastructure used by Netflix, Airbnb, and NASA.

Why AWS?

• Scalability: Can handle 7,000 stocks now, could handle 70,000 with minimal changes
• Reliability: AWS has 99.99% uptime guarantees (better than my laptop!)
• Cost-Effective: Pay-per-use model means I only pay for what I use (~$10/month for this project)
• Industry Standard: Learning AWS is a valuable career skill

AWS Services I Used:

Backend (Python)

All the data processing, scoring logic, and AI integration is written in Python. Why Python?

• Great libraries for finance (NumPy, Pandas)
• Easy to read and debug
• Native AWS Lambda support
• Cursor AI excels at Python

Frontend (Vanilla JavaScript)

The website is built with pure HTML, CSS, and JavaScript — no complex frameworks. This keeps it fast, easy to understand, and perfect for learning.

• Responsive design works on mobile and desktop
• PWA (Progressive Web App) support for installing on phones
• Chart.js for price history graphs
• Hosted on S3 + CloudFront for global CDN

Infrastructure as Code (Terraform)

Instead of manually clicking through AWS console, all infrastructure is defined in Terraform code. Benefits:

• Reproducible — can recreate entire system from code
• Version controlled — track changes over time
• Testable — deploy to test environment first
• Documented — code describes the architecture

Data Sources

Financial data comes from Alpha Vantage, a free API that provides:

• Company fundamentals (revenue, earnings, debt, etc.)
• Financial ratios (P/E, P/B, ROE, etc.)
• Price history (for charts)
• Covers 7,000+ US stocks

Key Takeaways

• You don't need to know how to code — AI tools like Cursor and Kiro can generate all the code from your descriptions
• But you do need to understand concepts — how databases work, what APIs are, how cloud infrastructure connects together
• Break big problems into small pieces — I didn't build this all at once; I started with one stock, then 10, then 7,000
• Ask lots of questions — the more specific your questions to the AI, the better the results

What I Learned About Investing

• Financial Statements: Balance sheets, income statements, and cash flow tell the story of a company's health
• Valuation Metrics: P/E ratios, P/B ratios, and ROE help determine if a stock is cheap or expensive
• Risk Indicators: Piotroski F-Score, Altman Z-Score, and Beneish M-Score help identify risky or potentially fraudulent companies

Challenge 1: API Rate Limits

Problem: Alpha Vantage free tier limits to 25 requests/day. I needed to fetch data for 7,000 stocks.

Solution: Implemented a tiered ingestion strategy — prioritize high market-cap stocks, add delays between requests, and spread the refresh over 24 hours.

Challenge 2: Data Inconsistencies

Problem: Some stocks had percentage values as decimals (0.15), others as whole numbers (15), causing wildly incorrect scores.

Solution: Built normalization functions to detect and standardize data formats before scoring. Added validation checks and confidence indicators.

Challenge 3: Cold Start Times

Problem: Lambda functions took 3-5 seconds to start when not recently used, making the API feel slow.

Solution: Added CloudFront caching for static assets, optimized Lambda package size, and implemented database query optimization with Global Secondary Indexes.

Challenge 4: Cost Control

Problem: AWS can get expensive if not managed carefully.

Solution: Used serverless architecture (pay-per-use), set billing alarms, optimized database queries, and implemented caching. Total cost: ~$10/month.

Short-Term (Next 3 Months)

• Add international stocks (UK, Europe, Asia)
• Implement user accounts and watchlists
• Add email alerts for score changes
• Mobile app (React Native)

Long-Term (Next Year)

• Backtesting engine to validate scoring methodology
• Machine learning to improve AI scoring accuracy
• Community features (share stock ideas, discuss)
• Premium tier with real-time updates

AI Changes Everything

A few years ago, building something like this would have required years of coding experience and a team of developers. Today, with AI "vibe coding" tools, a high school student can build a production-quality system in weeks.

This isn't just about stock analysis. It's about a future where anyone with an idea can build software — without needing to learn programming languages first.

The barrier to creating technology is disappearing. What will you build?