A Non-Cooperative Game-Theoretic Framework and Stochastic Predictive Matrix for Dynamic 72-Hour Airline Seat Capacity Liquidation

Abstract

This paper addresses a fundamental non-linear control optimization paradox within commercial aviation yield management: the systematic degradation of Expected Marginal Seat Revenue (EMSR-b) models under sudden demand shocks and high-frequency booking telemetry variances during the critical final 72-hour pre-departure window. This degradation traditionally results in significant perishable asset losses, leaving unallocated seat inventories across distribution streams. To resolve this inventory perishability dilemma, we introduce a non-cooperative game-theoretic framework driven by a real-time predictive stochastic matrix operator, designated as the Elasticity-Weighted Pricing (EWP) model. The proposed algorithm treats the continuous pre-flight liquidation window as a multi-player, non-zero-sum dynamic game between operating air carriers and aggregated passenger demand segments. By constructing a multi-dimensional capacity degradation matrix that weights temporal asset expiration against non-cooperative purchasing probability distributions, the model dynamically calculates a specialized Delta Pricing adjustment vector (ΔP). Crucially, to protect the baseline revenue integrity of legacy corporate booking channels, the architecture enforces a continuous boundary constraint using Karush-Kuhn-Tucker (KKT) complementary slackness conditions, achieving total Zero-Cannibalization Control (Z-Core).Extensive simulation-based backtesting utilizing synthetic route telemetry logs from 17 simulated carrier profiles based on industry-standard benchmarks validates a net revenue yield improvement boundary strictly localized between 5.5% and 18.4% on distressed cabin capacity, alongside a verified 11.2% median yield lift inside the continuous 72-hour pre-flight timeline.