This report explores tactics and game balance in Fighter Duel Lite, a two-player aerial combat simulation, using Monte Carlo Tree Search (MCTS). Our analysis revealed that move order and pilot expertise significantly influenced outcomes, with first-move advantage and specific opening maneuvers correlating with higher success rates. The analysis was done by developing an MCTS-based agent that employs some Fighter Duel Lite specific heuristics to guide the search, combined with a reward function to evaluate game states. The MCTS-agent achieved a win rate exceeding 90\% against a heuristically filtered random opponent, maintaining strong performance across varying starting positions and pilot skill configurations.

Den h¨ar rapporten unders¨oker taktiker ochspelbalans i Fighter Duel Lite, en tv˚aspelar-simulering av luft-strid, med hj¨alp av Monte Carlo Tree Search (MCTS). V˚aranalys visade att turordning och piloternas skicklighet hade storp˚averkan p˚a utfallen, d¨ar f¨orstadrag och vissa ¨oppningsman¨ovrarkorrelerade med h¨ogre vinstfrekvens. Analysen genomf¨ordesgenom att utveckla en MCTS-baserad agent som anv¨ander vissaspelanpassade heuristiker f¨or att styra s¨okningen, i kombinationmed en bel¨oningsfunktion f¨or att utv¨ardera spelst¨allningar.MCTS-agenten uppn˚adde en vinstfrekvens p˚a ¨over 90% moten heuristiskt filtrerad slumpm¨assig motst˚andare och bibeh¨ollstark prestanda ¨over varierande startpositioner och pilotkonfig-urationer.