CoinLand

The stock market run on Kid Logic, Finance made fun!

Comment

Inspiration

Let's be honest: Financial literacy is boring. Kids don't care about "interest rates," "quarterly earnings," or "diversified portfolios." They care about video games, movies, and designing their own space.

We realized that kids actually do understand market forces, they just don't know it yet. When the ice cream machine breaks at McDonald's, they know it's bad news. When a new Frozen movie is announced, they know it's huge. We built CoinLand to bridge that gap, teaching finance through the lens of "Kid Logic," connecting the abstract world of stocks to a tangible goal: Building the ultimate Dream Room.

What it does

CoinLand is a stock market simulator where financial wins turn into virtual rewards. The game loop is simple but educational:

  • Ride the Chaos: Users trade real stocks (Disney, Roblox, Tesla) based on Real-Time News Events generated by our AI engine.
  • "Tesla releases a mini-Cybertruck for toddlers" → TSLA goes UP 🚀

  • "Roblox servers crash during a live concert" → RBLX goes DOWN 📉

  • Earn & Plan: Smart trades earn "Coins." Kids learn to budget their profits, setting savings goals.

  • Rent or Buy: Users head to the Marketplace to furnish their personal room. They face real financial choices:

  • Buy a starter beanbag for cheap?

  • Rent a luxury gaming setup for the weekend?

  • Save for that rare neon sign?

  • Socialize: Visit friends' rooms, leave gifts, and get inspired by their designs.

The Lesson: Money isn't just numbers on a screen; it's a tool to build the life (or room) you want.

How we built it

We built a sophisticated quantitative trading engine disguised as a cozy room decorator.

  1. The "Newsroom" (Gemini AI): We used Google Gemini 2.0 Flash to generate 540 unique hours of financial news scenarios. We fine-tuned the prompt to ensure every headline was kid-friendly yet mathematically weighted with an "Impact Score" .
  2. The Backend (Python & Flask): Our custom engine acts as the market maker. It digests the news impact and calculates the new price based on sector volatility :

This ensures that "Tech" stocks are riskier/more volatile than "Food" stocks.

  1. The Ledger (Supabase): We used Supabase to manage not just stock portfolios, but also furniture inventory, rental timers, and room layouts, syncing the "Game State" across all players instantly via Realtime subscriptions.
  2. The Interface (React + Vite): A clean, gamified dashboard that seamlessly switches between the high-speed Stock Exchange and the cozy Room Designer.

Challenges we ran into

  • The "Too Creative" AI: At first, the AI wanted to write long, complex Wall Street Journal-style articles. We had to prompt-engineer it down to punchy, "Twitter-style" breaking news that a 10-year-old could digest in seconds.
  • Rate Limiting (Error 429): We hit the Google Gemini API quota pretty hard! To solve this, we built a Batch Generation System and a fallback "Infinite Loop Strategy" that ensures the game engine never crashes, even if the AI goes offline.
  • Balancing the Economy: We had to carefully tune the prices of furniture vs. stock returns. If stocks paid too little, nobody could afford a couch. If they paid too much, everyone bought the mansion on Day 1. We used a "Rent vs. Buy" mechanic to balance this—allowing kids to access expensive items temporarily, teaching cash flow management.

Accomplishments that we're proud of

  • The 540-Hour Engine: We successfully pre-generated enough dynamic content for 22.5 in-game days of continuous play without a single repeat.
  • Real-Time Math: The way our Python script calculates "Sector Volatility" feels genuinely realistic—users actually feel the stress when the market dips!
  • The "Rent vs. Buy" Lesson: Implementing a system where items can be rented added a layer of financial depth that most kid games miss.

What we learned

  • Prompt Engineering is an Art: Getting an AI to "think like a kid" required many iterations of tuning constraints and examples.
  • Database Design Matters: Structuring our Supabase tables to handle both "Financial Assets" (stocks) and "Physical Assets" (furniture) in a single user view was a great lesson in relational database design.
  • Gamification Works: We found that when you attach a boring concept (saving money) to a fun goal (buying a cool rug), users naturally want to learn the boring part to get the fun reward.

What's next for CoinLand

  • Co-Living Spaces: Pooling money with friends to rent a "Clubhouse" together.
  • More Sectors: Adding "Sneakers" (Nike/Crocs) and "Candy" (Hershey's) to the market to expand the economy.
Share this project:

Updates