Counter-Strike Desk — Loading…

HLTV-style ELO (1000-1800 scale) per team. Series win-prob from binomial best-of-N. Self-learning ELO updates as match scores arrive.

P(sweep) = p^target. P(distance) = decisive last game. Map total OVER 2.5/4.5 based on series length.

• HLTV-style 1000-1800 ELO scale (Vitality 1740, MOUZ 1720 baseline)
• K-factor 20, log-scaled by map margin
• Series win prob via binomial best-of-N
• Liquipedia upcoming matches fed as primary source; top-30 cross-pairs as fallback
• Self-learning: each completed match updates ELO → next prediction