Hollywood Stocker

Creating a Real-Time Stock Market Sim for Hollywood Actors

Case Study

If you played Hollywood Stock Exchange in the early 2000s, you know what's up. It was awesome. "The Town" podcast also does a HSX segment every year, so the last time they did it, it got me all fired up to invest in some actors. I went to revisit the OG HSX recently and, honestly, the site was kind of a mess. So I vibe coded my own with Cursor!

Hollywood Stocker is a sophisticated stock market simulation that treats Hollywood actors as tradable assets. I directed AI agents to assemble a fully functional trading platform where an actor’s StarScore (aka share price) is calculated from real Wikipedia page views, TMDb popularity scores, and acting credits. The market updates automatically twice daily (12am and 12pm) with fresh data, creating an authentic trading experience with realistic market dynamics, professional analytics, and advanced sell strategies straight from real trading playbooks.

Goals

What I Learned

I knew exactly what I wanted to create when I started Hollywood Stocker. As the backend evolved, I refined the frontend—and when new ideas hit the frontend, I’d circle back to the backend. It was a true iterative process that taught me how to direct something from the ground up.

When things broke, I had my Cursor agents investigate, but I quickly realized that most of the real breakthroughs came from me. Even as a beginner, I was the one diagnosing the core issues, defining what needed fixing, and steering the AI in the right direction.

I didn’t learn how to code every detail—like parsing “Lupita Nyong’o” vs. “Lupita Nyong%27o”—but I learned how to spot those problems and communicate exactly what needed to be solved.

I wanted market movements that felt organic, not random. I didn’t code the 30-day rolling window for Wikipedia views myself, but I figured out that it was the key to creating realistic bull runs and corrections—and directed Cursor to wire the logic around that.

Eventually, when the cron job kept failing, I ran a full system audit with multiple agents. They all agreed we needed file locking, stale lock detection, progress tracking, graceful error recovery, and batch processing with rate-limit delays. At first, I didn’t fully grasp those terms—but now I understand exactly how they work together to keep the system stable.

What I Directed

StarScore Calculation System

The heart of Hollywood Stocker is the StarScore — a weighted formula combining three real-world data sources with scaling factors to normalize the data into a tradable price:

StarScore = (Wikipedia/1000 × 0.60) + (TMDb/10 × 0.25) + (Credits/10 × 0.15)

Is this the most perfect formula to gauge an actor's real-world market value? Absolutely not. But it's incredibly accurate for three open source APIs that are updated daily.

Breaking Down the Formula:

• Wikipedia Page Views (60% weight): 30-day rolling total divided by 1,000 to scale down massive view counts into manageable numbers. A viral actor with 100,000 views contributes 60 points to their StarScore.
• TMDb Popularity (25% weight): Movie database popularity score (0-100 scale) divided by 10, then weighted. An A-lister with TMDb 85 contributes 2.125 points.
• Acting Credits (15% weight): Total professional roles divided by 10 to prevent veterans with 200+ credits from skewing the market. An actor with 45 credits contributes 0.675 points.

Real Example:

An established star with 50,000 Wikipedia views, TMDb score of 85, and 45 acting credits:

• Wikipedia: (50,000 / 1,000) × 0.60 = 50 × 0.60 = 30.0
• TMDb: (85 / 10) × 0.25 = 8.5 × 0.25 = 2.125
• Credits: (45 / 10) × 0.15 = 4.5 × 0.15 = 0.675
• Final StarScore: $32.80 per share

Actor Classification System

Every actor gets classified into stock categories with visual indicators:

Stock Categories

• 🔵 Blue-Chip ($80+): A-list superstars with decades of reliable performance
• 🟢 Green-Chip ($30-79): Established stars with proven track records
• 🟡 Mid-Cap ($5-29): Rising stars on a steady climb
• 🔴 Penny Stock ($0-4.99): Unknown newcomers or character actors

Performance Classifications

The system also tracks performance patterns to identify trading opportunities:

Full Trading Platform

The system supports complete portfolio management: buy shares at current prices, sell partial or full positions, automatic weighted-average cost calculation for multiple purchases, and a complete transaction history stored in the Ghostnet ledger. Cash management is unified across my entire site ecosystem — dollars spent here affect your balance everywhere.

Sell Strategy Analysis

I directed the rollout of ten professional trading strategies that analyze every position in real-time. Each strategy shows color-coded status indicators (✅ SELL, ⚠️ WATCH, ❌ HOLD, 🎯 COMPLETE) telling you exactly when to take action:

• Partial Profit Rule: Sell 50% when you're up 25%+ and momentum declines
• Double & Dump: Exit when your investment doubles
• Trailing Stop Loss: Sell if price drops 10% from peak
• Wikipedia Spike Exhaustion: Sell when viral moments roll out of the 30-day window
• RSI Fade, Moving Average Crossover, Support Break, and three others covering momentum indicators, technical analysis, and divergence patterns

Sell Strategy Analysis: Professional sell strategy analysis for every position.

Portfolio Features

Automated Data Pipeline

The system currently runs twice daily at 12:01 AM and 12:01 PM, keeping prices fresh. The update process is sophisticated and bulletproof.

Data Processing

Intelligent Fallbacks

Monitoring & Logging

The whole system runs itself — I wake up to fresh market data, check prices again in the afternoon, and if something goes wrong, the comprehensive logging makes debugging straightforward.

The Tech Stack

Frontend

• 13,757 Lines of Vanilla JavaScript: ChatGPT is obsessd with putting the number of lines of code in these case studies, so HERE YOU GO! No frameworks — data loading, rendering, analytics calculations, chart building, search system, and backup functionality
• Chart.js + Financial Extension: Interactive line and candlestick charts with custom date range pickers and time range filters. One of the main reasons I created this project was so I could have some kick ass candlestick charts

The codebase is modular and scalable — adding new features is straightforward because everything is organized into clean, reusable functions, with exhaustive markdown documentation.

When something breaks, I give Cursor the markdown file and the error message, hit enter, and about 15 minutes later (after quite a bit of back and forth - and some cursing), not only is the code fixed, but I usually receive a huge system upgrade that requires me to make changes to this very page.

Backend & APIs

• PHP Backend: Trading endpoints (buy/sell), data validation, and JSON file-based storage for portfolio, history, and transaction ledger
• TMDb API: Actor search, popularity scores, filmography data, and profile photos
• Wikipedia Pageviews API: 30-day rolling page view totals with historical data access and intelligent fallback system

Automation & Integration

The cron system runs autonomously with zero intervention — batch processing respects rate limits, lock files prevent conflicts, and comprehensive logging catches every edge case. Ghostnet integration means Hollywood Stocker isn't standalone — it's part of a unified financial ecosystem across the entire site.

Why It Rocks

The market dynamics are real. Once the game launched, I had no idea if the StarScore or API data would be accurate or even make sense. But it turns out, it's pretty damn accurate! Here are a few examples of the market dynamics in action:

Additional Challenges

• API Rate Limiting: TMDb has strict limits — learned to batch requests, implement smart caching, and use exponential backoff. The asset cache alone saves thousands of calls.
• State Management Complexity: 13K+ lines managing portfolio data, history arrays, ledgers, chart states, and UI updates taught me why frameworks exist — but doing it in vanilla JS taught me how state actually works.
• Tracking Sold Actors: Originally, sold actors stopped updating. Built a system that parses the Ghostnet ledger to continue tracking their StarScore for historical comparison.
• Mobile Responsiveness: Making complex financial charts, expandable actor cards, and strategy panels work smoothly on mobile took serious CSS and JS work. This will be the process of Hollywood Stocker I will continue to iterate on until I die.

Future Enhancements

Platform Expansion

• Build a native mobile app version with push notifications for sell strategy alerts
• Integrate real-time price updates using websockets instead of page refreshes
• Add social features: leaderboards, shared portfolios, and competition modes

Feature Depth

• Expand charting with advanced technical indicators (Bollinger Bands, MACD, Fibonacci retracements)
• Integrate dynamic LLM sell strategy analysis for each position
• Explore machine learning models to predict StarScore trajectories based on release schedules and trending patterns

Final Word

The Hollywood Stock Exchange has a special place in my heart. I used to play it with my mom, and I've wanted to make my own version of it for years. Hollywood Stocker was a fun and challenging project that taught me that, in the age of "vibe coding," if you can dream it, it's possible to direct an AI tool to make it.