ICEC 2024 · Springer LNCS 15192 · first author
Dungeon Wipe: Exploring Dynamic Difficulty Adjustment with Power-Up Mechanics
The game reads health thresholds to decide potion support and leaves enemy count, damage and behaviour alone. A Monte Carlo simulation over 5 levels at 1,000 runs each put modelled failure at 8% with adaptive support, against 40% and 56% for the fixed distributions. Of 38 computer-science students who played it, 83% did not perceive the adaptation.
Read publication↗ (opens in a new tab)Master’s thesis in progress, not yet submitted, expected 2026
I map Riot’s stated balance intentions onto champion-specific metrics across patches 25.20 and 25.21, split by rank, role and match phase. The current analysis covers 3,362 participant observations. Of the 12 champions classified so far, six aligned with the stated intention and five did not.