4th Edition of Psychiatry and Addiction World Conference 2026

Abstract

Loss chasing is widely recognised as a core feature of gambling disorder, yet its underlying mechanisms remain poorly understood. This presentation reports a programme of five field studies conducted with Nilosmita Banerjee and Zhang Chen, examining loss chasing at scale in ecological conditions, across more than 170 million rounds of play and over 11,000 gamblers drawn from a real online gambling operator. A preliminary scoping review of 74 studies established that loss chasing is not a unitary construct: persistence, stake changes, and speed of play represent distinct behavioural facets driven by different mechanisms. Four empirical studies then investigated these facets within and between sessions, using dice, slots, blackjack, and roulette data. Within sessions, post-loss acceleration of play emerged as the most robust and theoretically pure indicator of chasing: unlike stake adjustment, speed of play is unconstrained by financial resources and reflects the motivational urge to continue directly. Between sessions, gamblers returned faster and wagered more after winning sessions, consistent with wealth and house money effects. Critically, large losses also produced faster returns and increased wager intensity in the subsequent session, a break even effect that contradicts laboratory predictions grounded in the gambler’s fallacy and underscores the primacy of financial realities in field contexts. Slot gamblers and high involvement gamblers showed outcome insensitivity across indicators, a pattern consistent with habituation and dark flow immersion. Across the five studies, involvement level moderated effects in a theoretically revealing way: high involvement gamblers were more sensitive to outcome magnitude but less sensitive to win versus loss direction, suggesting that escalating engagement is accompanied by a progressive decoupling of behaviour from individual outcomes. These findings carry direct implications for operator level detection: no single marker is sufficient, and a complementary trio combining speed of play, losing streak persistence, and time of return offers broader coverage of the risk space. The programme illustrates the unique empirical leverage that large scale online data provides, while acknowledging its inherent limitations: single operator scope, absence of validated clinical screening, and no direct access to account balance trajectories.