HiddenBag Discord → Supabase Rework Plan

Date: 2026-02-09

Current Architecture (what exists)

• Telegram channels → FULL-LOCAL-PRODUCTION collectors → message_queue → Discord forwarders
• Sheets → sheets-to-supabase.py → hb_picks table
• hb_picks → consensus SQL → hb_consensus_snapshots → grading
• 79 cappers tracked, 41 free, 2 algo source
• Discord server for aggregation: 1469086083701542956

Tables in Supabase:

• hb_cappers (name, slug, tier, sports, aliases)
• hb_picks (capper_id, sport, team, line, odds, units, result, raw_text)
• hb_discord_messages (raw discord messages)
• hb_game_odds (game lines/scores for grading)
• hb_consensus_snapshots (consensus plays)
• team_mappings (team name normalization)
• hb_capper_stats (materialized view)

New Architecture (Matt's request):

• Discord server 1469086083701542956 as primary data source
• Each channel → tracked in Supabase
• Some channels have multiple cappers (need splitting)
• Still pull some data from Google Sheets
• CSV export for validation
• Cleaner data, better consensus