Quantifying player death impact in League of Legends

Ferreira, Ruben (avtor), Faganeli Pucer, Jana (avtor)

League of Legends is one of the world’s most played and watched e-sports. We study how the death of an individual player affects team win probability. From Diamond-tier matches, we construct minute-level snapshots to train a calibrated win-probability model. We formalize a death window based on the state just before death until the first state after the respawn. We attribute the team-level win-probability change to players via a role/time standardized Performance Score weighted by SHAP feature scores. Results across 150,472 death windows show, the team of the deceased player loses on average 2.03% of win probability; probability decreases in 58% of death windows and increases in 42%. The approach enables fast identification of impactful deaths and targeted learning opportunities.

• Jezik: Angleški jezik
• Ključne besede: esports analytics, win probability, player performance attribution, League of Legends
• Tipologija: 1.08 - Objavljeni znanstveni prispevek na konferenci
• Organizacija: FRI - Fakulteta za računalništvo in informatiko
• Status publikacije: Objavljeno
• Različica publikacije: Objavljena publikacija
• Leto izida: 2025
• Št. strani: Str. 51-54
• PID: 20.500.12556/RUL-177555
• UDK: 004.9
• COBISS.SI-ID: 262492419
• Datum objave v RUL: 24.12.2025

Gradivo je del monografije

• Naslov: Proceedings of the 11th Student Computing Research Symposium : (SCORES'25)
• Uredniki: Uroš Čibej, Luka Fürst, Lovro Šubelj, Jure Žabkar
• Kraj izida: Ljubljana
• Založnik: Založba UL FRI, = Faculty of Computer and Information Science
• Leto izida: 2025
• ISBN: 978-961-7059-18-2
• COBISS.SI-ID: 261672195

Licence

• Licenca: CC BY-SA 4.0, Creative Commons Priznanje avtorstva-Deljenje pod enakimi pogoji 4.0 Mednarodna
• Povezava: http://creativecommons.org/licenses/by-sa/4.0/deed.sl
• Opis: Ta licenca Creative Commons je zelo podobna običajni licenci Priznanje avtorstva, vendar zahteva, da so materialne avtorske pravice na izpeljanih delih upravljane z enako licenco.

Sekundarni jezik

• Jezik: Slovenski jezik
• Ključne besede: analitika e-športa, verjetnost zmage, pripisovanje uspešnosti igralcev, League of Legends