A data analyst systematically evaluated the betting predictions of three Russian football pundits on La Liga matches across 26 matchdays (259 predictions) in the 2025-2026 season. Using R and LLMs (Gemini and Claude) to parse screenshots from YouTube videos, the author compiled a dataset including predicted winners, odds, and actual outcomes. Results show that even deep domain knowledge doesn't translate to profitable betting — two of three pundits won 44% of bets, landing near a neutral balance. Interestingly, some pundits showed strong team-specific accuracy (e.g., Lukomskiy at 66.7% on Villarreal, Grabchak at 7/7 on Osasuna), while all three consistently failed on the big three clubs (Real Madrid, Barcelona, Atletico Madrid).
1.5K Impressions