A computer system, a computer program product, and a computer-implemented system are provided for selectively restricting fiscal transactions based on peer-to-peer (P2P) interaction events. NLP is leveraged to detect a future P2P interaction event, and historical records are leveraged to detect and use historical P2P interaction events and historical financial transactions to predict a fiscally related expenditure for the future P2P interaction event. Depending upon whether the predicted expenditure exceeds a spending control limit, action is taken to restrict the fiscal transaction(s).