🛍ïļ Retail Forecaster: Black Friday Edition

Master Time Series Forecasting Through Real Retail Chaos

⏱ïļ 30-45 minutes
ðŸ‘Ĩ Teams of 2-3
ðŸŽŊ Real-world data

ðŸŽŪ Game Overview

Welcome to the most intense week in retail! You're the inventory manager for "TechMart," a consumer electronics retailer. Your mission: predict demand for Black Friday week without drowning in excess inventory or facing angry customers with empty shelves.

📊 Your Data Arsenal

  • 3 years of daily sales history
  • Product categories: TVs, Laptops, Gaming, Smart Home
  • External factors: weather, competitor ads, social media buzz
  • Historical Black Friday patterns

ðŸŽŊ Your Mission

  • Forecast daily demand for Black Friday week
  • Optimize inventory levels
  • Minimize costs while maximizing revenue
  • Handle surprise events in real-time

⚡ The Challenges

  • Storage costs: $5 per unit per day
  • Stockout penalty: $50 per missed sale
  • Rush shipping: $20 per unit
  • Customer satisfaction metrics

📈 Game Progression

Round 1
Data Exploration
(10 min)
Round 2
Model Building
(15 min)
Round 3
Initial Forecast
(10 min)
Round 4
Black Swan Event
(5 min)
Round 5
Final Results
(5 min)

ðŸŽē Round Details

Round 1: Data Exploration (10 minutes)

Teams receive historical sales data and must identify:

  • Overall trends (growing? declining?)
  • Seasonal patterns (weekly, monthly, yearly)
  • Previous Black Friday spikes
  • Product correlations
# Sample Data Preview Date | Product | Units_Sold | Avg_Price | Weather | Promo ------------|------------|------------|-----------|---------|-------- 2022-11-25 | TV-55inch | 145 | $399 | Clear | Yes 2022-11-26 | TV-55inch | 98 | $449 | Clear | No 2022-11-27 | TV-55inch | 67 | $449 | Rain | No ...

Round 2: Model Building (15 minutes)

Choose your forecasting approach:

Available Methods:
  • Simple: Moving averages, trend projection
  • Seasonal: Decomposition, seasonal indices
  • Advanced: ARIMA, exponential smoothing
  • External: Include weather, promotions, social media

Round 3: Initial Forecast (10 minutes)

Submit predictions for Black Friday week:

  • Daily forecasts for each product category
  • Confidence intervals
  • Inventory order quantities
  • Contingency stock levels

Round 4: Black Swan Event! (5 minutes)

BREAKING: Your competitor just announced bankruptcy! Their customers are flooding to your store. Social media is going crazy. You have 5 minutes to adjust your forecasts!

Quick decisions needed:

  • Rush order more inventory? (costs extra!)
  • Raise prices to manage demand?
  • Limit quantities per customer?

Round 5: Final Results (5 minutes)

See how your predictions performed against actual sales!

📊 Scoring System

Metric Calculation Points
Forecast Accuracy 100 - MAPE (Mean Absolute Percentage Error) Up to 400 points
Inventory Efficiency Revenue - (Storage Costs + Stockout Penalties) Up to 300 points
Customer Satisfaction % of demand fulfilled without delays Up to 200 points
Black Swan Response Speed and effectiveness of adjustment Up to 100 points
Total Possible 1000 points
Profit Calculation:
Profit = (Units_Sold × Price) - (Excess_Inventory × $5/day) - (Stockouts × $50) - (Rush_Orders × $20)

🏆 Leaderboard Categories

Awards to Win:

  • ðŸŽŊ The Oracle: Highest forecast accuracy
  • 💰 The Profit Maximizer: Highest total profit
  • 😊 Customer Champion: Best satisfaction score
  • ðŸĶĒ Black Swan Survivor: Best crisis response
  • 📈 Most Improved: Biggest gain from Round 1 to Round 5

ðŸ’Ą Strategy Tips

Rookie Mistakes to Avoid

  • Ignoring day-of-week patterns
  • Over-fitting to last year only
  • Missing the post-Thanksgiving surge
  • Not accounting for product cannibalization

Pro Strategies

  • Weight recent Black Fridays more heavily
  • Build separate models for different products
  • Include leading indicators (search trends)
  • Keep safety stock for top sellers

Data Patterns to Find

  • Pre-Black Friday dip (people waiting)
  • Cyber Monday online spike
  • Weather impact on foot traffic
  • Social media buzz correlation

🎓 Learning Objectives

By playing this game, students will:
  • Experience the complexity of real-world forecasting with messy data
  • Understand the trade-off between over-stocking and under-stocking
  • Learn to identify and model multiple seasonalities
  • Practice adjusting models when conditions change rapidly
  • See how external factors affect time series predictions
  • Develop intuition for when simple vs. complex models work best

🚀 Implementation Options

📝 Debrief Questions

After the game, discuss these key points:

  1. What patterns did you notice first? Which ones did you miss?
  2. How did your forecast change when you added external variables?
  3. What would you do differently with more time or data?
  4. How did the Black Swan event expose weaknesses in your model?
  5. Which was more costly: over-forecasting or under-forecasting? Why?
  6. How would this apply to other industries (healthcare, airlines, restaurants)?