Currently, Amazon relies on third parties for transportation marketplace rate forecasts, despite the poor quality and lack of interpretability of these forecasts. While transportation marketplace rates are typically very challenging to forecast accurately, we have developed a novel signature-based statistical technique to address these challenges and built a predictive and adaptive model to forecast marketplace rates. This novel technique is based on two key properties of the signature transform. The first is its universal nonlinearity which linearizes the feature space and hence translates the forecasting problem into a linear regression analysis; the second is the signature kernel which allows for comparing computationally efficiently similarities between time series data. Combined, these properties allow for efficient feature generation and more precise identification of seasonality and regime switching in the forecasting process. Preliminary result by the model shows that this new technique leads to far superior forecast accuracy versus commercially available industry models with better interpretability, even during the period of Covid-19 and with the sudden onset of the Ukraine war.