Skip to main content
CCU Forecasting uses Meta’s Prophet time-series model to predict future player counts for any tracked game.

How It Works

  1. Historical CCU data is collected every 5 minutes
  2. Data is aggregated into daily or hourly buckets
  3. Prophet model is trained daily (4 AM) on the aggregated data
  4. Forecasts project 7–30 days into the future with confidence intervals

Features

  • Point forecasts: Expected CCU for each future day
  • Confidence intervals: Upper and lower bounds showing prediction uncertainty
  • Trend decomposition: See seasonal patterns (weekday vs. weekend, holidays)
  • Multiple horizons: Short-term (7d) and long-term (30d) projections

API Endpoint

GET /api/forecast/{place_id}
See API Reference → Forecast for full details.

Plan Limits

PlanForecast Access
Explorer
BuilderBasic (7-day)
StudioAdvanced (30-day + confidence)