Advantage play bots were silently sapping 30% of profits every minute of everyday

— Major live dealer iGaming operator

Bar chart showing percentage of win rate on turnover or wager for an online operator before and after implementation of an improvement. The red bar represents the pre-implementation rate, and the two blue bars represent the rate after the implementation, with a 35% increase indicated.

Engagement Differential Labs was engaged by a major iGaming operator who derived the majority of revenue from live table games. This operator had persistently low win rates (GGR divided turnover).

What did we do Using AI driven methods, we discovered a group of players exhibiting unusual play behaviors and behaviors consistent with computerized play. On inspection, these unusual actors all had volume based marketing rebates of ~.8% of turnover. Through optimal betting these unusual players reduced their edge to .35% of turnover meaning once rebates are considered these players had a consistent edge over the house.

Impact The Differential Team worked with the operator to harden dealing procedures and root out nefarious actor. Immediately after implementing safeguards the win rate on turnover increased from less than 1% to the theoretical edge of 1.29% (see chart at right).

About the proliferation of advantage play

Screen capture of CNA news website with headline about a syndicate member jailed for using a secret baccarat formula at MBS casino in Singapore, and an image of a casino floor with slot machines.

We have been significantly involved in the proliferation of side-bets across Asia. While these bets can be beneficial they also are often countable and can even be exploited with a deep cut card. Some examples of these risks,
- Multiple teams have been identified in Macau over the last three years
- Teams have been arrested and prosecuted in Singapore
- In Vegas BJ count teams have migrated to target baccarat side-bet

Traditional combinatorics based mathematics is inadequate to protect side-bets with complex rules. We have developed a machine learning approach that identify patron opportunities and flag patrons exploiting these weaknesses.