Player Behavior

Active users, churn, new vs returning, sessions per player and behavioral analytics on the Atlas platform.

Player Behavior Analytics covers the player-engagement layer of the Atlas Intelligence module — active users, churn rolling D7/D14/D30, new vs returning per month, sessions per player and weekly heatmap by hour. Powered by all event types and snapshotted daily in ClickHouse.

Available Features

Active Users (live)

• DAU / WAU / MAU — Unique players in window, computed via uniqHLL12 over Gold tables. Available in /v1/analytics/casino/overview (players field) and /v1/analytics/casino/players-summary (total_players).

Retention & Churn (live)

• Churn D7 / D14 / D30 — Rolling churn flags per player based on days since last bet. Available in /v1/analytics/casino/players-summary. Snapshot refreshed daily; snapshot_age_days exposed in payload to flag stale data.
• New vs Returning — Monthly breakdown of first-time vs recurrent players for up to 12 months. Available in /v1/analytics/casino/new-vs-returning.
• Cohort Analysis — Planned: retention curves by registration cohort.
• Reactivation Tracking — Planned: win-back campaign effectiveness.

Engagement (live)

• Sessions per Player — Average session count per active player. Available in /v1/analytics/casino/players-summary (sessions_per_player).
• Volume by Hour — 24-bucket BRT distribution of stake/GGR per hour, with optional bucket filter (slots/live/tables/outros). Available in /v1/analytics/casino/volume-by-hour.
• Weekly GGR Heatmap — 7×24 matrix (day-of-week × hour BRT) of GGR with min/max for gradient rendering. Available in /v1/analytics/casino/weekly-heatmap.

Player Segmentation

• RFM Analysis — Planned: Recency, Frequency, Monetary value segments.
• VIP Tiers — Planned: automatic classification based on betting volume.

LTV Modeling

• Predicted LTV — Planned: AI-powered lifetime value prediction.
• LTV by Acquisition Channel — Planned.
• Segment LTV — Planned.

Endpoints

Path	Returns
/v1/analytics/casino/players-summary	Total/Novos/Recorrentes + Sessions/Player + Churn D7/D14/D30 + snapshot_age_days
/v1/analytics/casino/new-vs-returning	Monthly breakdown (event_month, new_players, returning_players, total_players)
/v1/analytics/casino/volume-by-hour	24 buckets BRT (hour, stake, ggr, hold_pct, rounds)
/v1/analytics/casino/weekly-heatmap	7×24 matrix + min/max GGR
/v1/analytics/sport/players	Sport-side player activity

Data Prerequisites

Player behavior analytics requires all event types:

• User events — Registration and profile data
• Transaction events — Deposit/withdrawal patterns
• Casino events — Game preferences (casino_session_id recommended for accurate session counts)
• Sports events — Betting patterns

Churn rolling depends on a daily snapshot job. In staging/production this is handled automatically by a Refreshable Materialized View. Local dev requires manual npm run refresh:churn in the data/ module.