International merchandise trade statistics are closely monitored in the ASEAN+3 due to the importance of trade on the region’s economies. However, statistics are typically released with a lag of at least one month, creating an information gap for real-time policy decisions. To address this challenge, we present two frameworks for nowcasting export growth across the 14 ASEAN+3 economies. Firstly, bridge models integrate two indicators derived from ship traffic with a few financial variables, estimated via linear regression and machine learning (ML) techniques. Secondly, large-scale ML models address the risk of overlooking critical predictors with the use of over 100 external and domestic variables. While large-scale ML models generally show greater predictive power, this difference is not significant across all economies. Indonesia, Japan, Lao PDR, Malaysia, and Singapore clearly benefit from the larger ML models, while the simpler bridge models suffice for others. The large-scale ML models exhibit reasonable accuracy up to three months ahead. Their wide range of predictors can also compensate for the absence of ship traffic data in most cases.