Exploiting Arbitrage Opportunities in Live Sports Betting: An Automated Approach

Abstract

This thesis explores the existence and feasibility of arbitrage opportunities in live sports betting markets by analyzing odds from two major sportsbooks—FanDuel and BetMGM—across multiple NBA games. Arbitrage is defined as a risk-free profit opportunity that occurs when betting on both sides of a game across platforms result in a positive return. Using real-time data collected through web scraping, the study identifies moments when the sum of opposing moneyline odds exceeds zero, signaling a profitable opportunity. The results show that while arbitrage opportunities do occur, they are rare and short-lived, typically lasting around 13 seconds under favorable conditions. On average, arbitrage was present in only 4.52% of total scraped game time. Arb1 configurations—pairing FanDuel’s Team 1 odds with BetMGM’s Team 2 odds—were more frequent than Arb2,the opposite pairing, likely due to structural team ordering and sportsbook pricing tendencies. Locking behavior was also analyzed, revealing that sportsbooks occasionally freeze odds, but these lock events were generally not correlated with arbitrage instances. The thesis also investigates cross-state arbitrage by comparing odds both between sportsbooks across states (e.g., FanDuel in New Jersey vs. BetMGM in Indiana) and within the same sportsbook across states (e.g., BetMGM in New Jersey vs. BetMGM in Indiana). While some differences were observed—particularly on BetMGM—no arbitrage opportunities emerged. Moreover, even if such opportunities existed, they would be impractical to exploit due to geolocation restrictions and the inability to place bets in two states simultaneously. An algorithm was developed to detect and execute arbitrage in real time. While it successfully placed a test bet under controlled conditions, practical challenges—such as reloading betslips, site restrictions, and account risk—limit the scalability of automated arbitrage betting in live markets.