Gamifying the Stock Market

Turning a bar into a living financial market to make economics fun and accessible

Idea

The idea was born when one of our group discovered "Stock Market Bars" in Germany — places where drink prices fluctuate based on real-time demand, just like stocks.

We saw a unique opportunity: bring the stock market to Romania in a familiar, social setting.
Instead of complex charts and intimidating terminology, people would learn financial concepts while having a beer with friends.

How It Works

1. The bar offers a fixed set of products (beer, wine, soft drinks, etc.)
2. Prices change dynamically every 15 minutes based on actual purchases
3. Popular items become more expensive, less popular ones get cheaper
4. This creates a real supply-and-demand environment in a relaxed social setting

Goal

Financial education in Romania ranks among the lowest in the EU.

Our mission is to change that — not through lectures, but through experience.
By turning a bar into a living market, people can actively participate, make decisions, see consequences, and naturally understand core economic principles like supply and demand, market sentiment, risk, and crisis management.

Beyond the bar itself, we plan to host educational events and talks where guests can learn about real investing in an engaging way.

Technology

The system is built with a modern, scalable full-stack architecture.

• Frontend: TypeScript/React application that displays live prices and allows authorized users to manage the system.
• Backend: Python pricing engine that calculates new prices using an Exponential Moving Average (EMA) model.
• Database: MySQL for storing raw sales data.
• Storage: S3 for price history and the final aggregated live_prices.json consumed by the frontend.
• Infrastructure: Deployed on Railway with separate services for the UI and automated pricing cycles.

Algorithm

Core Concept

Traditional stock markets move at high frequency. In a bar, purchase frequency is much lower and more social.
To make the experience educational and enjoyable, we update prices every 15 minutes based on actual demand.

Key Variables

Base Parameters

• Base Prices and Bottle Costs
• Risk controls: Alpha per Share, Absolute Impact Weight, Depth, Maximum Change, Rounding (to 0.50), Soft Floor Delta

EMA & Smoothing

• Lambda & Lambda per Share
• Warmup Period
• Lambda Ramp

Crisis Management

• Crisis Mode (spike_only)
• Spike Multiplier

State Variables

The engine maintains memory of market behavior:

• Current Prices
• EMA Shares & EMA Volumes
• Warmup Count
• Crisis Triggered flag + Crisis Index
• Cumulative Profit & Cumulative Sold
• Event Label ("CRISIS", "SMOOTHED", etc.)

Algorithm Engine

1. Exponential Moving Average (EMA)

We use two EMAs to simulate realistic market movement:

• EMA on Volumes — tracks recent sales volume of each product
• EMA on Shares — measures relative popularity among products

Formula:

EMA_t = α × Value_t + (1 - α) × EMA_t-1

During the warmup phase the algorithm learns the market gradually. After warmup, it becomes more responsive to changes in demand.

2. Price Calculation

For each drink, the new price is calculated based on:

Current vs Expected market share (relative popularity) Current vs Expected volume (sales surprise) Dynamic depth and absolute impact weight (to prevent extreme swings)

The final price is then clipped, rounded to the nearest 0.50, and protected by a soft floor.

3. Crisis Detection

The system currently operates in spike_only mode. If any product’s price exceeds Base Price × Spike Multiplier, a crisis is triggered and all prices are reset to base levels to restore market balance and prevent unrealistic spikes.