Skip to main content

Prior Win/Loss - Horse Racing

Blair Matheson
  • Updated

The Prior Win/Loss - Horse Racing Version feature enables you to deliver highly targeted campaigns by notifying customers when horses they have previously interacted with (Won, Lost, or Bet) are racing again. This approach leverages past behaviors, whether celebrating a win or chasing a loss, to drive renewed engagement.

What it Does

  • Identifies horses users have previously bet on (Winner, Loser, Any prior bet).
  • Monitors upcoming popular races featuring those horses.
  • Skips users who already bet that specific upcoming race (dedupe).
  • Prioritizes recommendations based on user affinity and current race popularity.

How it Works

  1. Track Behavior: Uses horse-level prior behavior (win/loss/any) with recency/score.
  2. Monitor Inventory: Scans upcoming races within the time window that meet a minimum popularity threshold.
  3. Match: Finds upcoming races where the user’s horse is running.
  4. Filter: Applies odds and market/class/type filters.
  5. Dedupe: Excludes users who already placed a bet on that race in the last 2 days.
  6. Rank: Picks the top horse per user by score, and the best-priced (lowest odds within range) selection.

Configurable Parameters

  • Audience Type (audience_type):
    • 'Winner'
    • 'Loser'
    • 'Any Prior Winner'
    • 'Any Prior Bet'
  • Timing:
    • min_hours_before_race
    • max_hours_before_race
  • Quality & Filters:
    • min_popularity: Minimum unique bettors on the race.
    • min_proposed_odds / max_proposed_odds: Decimal odds range.
    • market_names / class_names / type_names: Whitelists (or 'Any').

Output Data

The following data fields are available in the output:

  • userid
  • horse_name
  • selection_key
  • track_name (aliased from event_name)
  • race_number (aliased from event_key)
  • post_time (aliased from event_date_time)
  • odds
  • market_name, market_key
  • type_name, type_key

Context (Historical):

  • last_event_name
  • last_event_date_time
  • last_market_name
  • last_odds
  • last_bet_potential_payout

Required Events and Tables from Client Integration

Event Stream:

  • Key fields: event_key, userid, brandkey, category_name, bet_action, bet_action_time, and date partitions (year, month, day).
  • Requires BET_PLACE and BET_SETTLED events.

Inventory Snapshot:

  • Key fields: eventdatetime, eventname, eventkey, categoryname, selectionname, selectionkey, marketname, marketkey, typename, typekey, classname, denprice, numprice, decprice.
  • Filters enforce category = horse racing, odds range, and optional market/class/type whitelists.

Related articles