The Temporally and Causally Adaptive Reinforcement Portfolio (TCARP) strategy presents a novel approach to algorithmic trading by integrating causal inference with reinforcement learning. Financial markets exhibit complex, non-stationary dynamics that challenge traditional trading algorithms and black-box machine learning approaches. TCARP addresses these challenges by leveraging causal discovery to identify genuine market drivers, which then inform a reinforcement learning agent’s decision-making process. The strategy features a temporal adaptation mechanism that detects regime changes and updates the causal understanding accordingly, enabling resilience to shifting market conditions. Additionally, TCARP incorporates an explainability layer that traces trading decisions through causal pathways, providing interpretable justifications for actions taken. Our preliminary experiments demonstrate that TCARP achieves competitive performance compared to baseline methods while offering superior adaptability to market regime shifts and transparent decision-making. These results suggest that combining causal inference with reinforcement learning represents a promising direction for developing more robust and interpretable algorithmic trading systems.