Context modifies the influence of any trading indicator. Ceteris paribus, a buyer would be more cautious buying in a selling market context than in a buying market. In order for automated, adaptive systems like neural networks to better emulate and assist human decision-making, they need to be context sensitive. Most prior research applying neural networks to trading decision support systems neglected to extract contextual cues, rendering the systems blind to market conditions. This paper explores the theoretical development and quantitative evaluation of context sensitivity in a novel fast learning neural network architecture, Echo ARTMAP. The simulated risk and cost adjusted trading results compare very favorably on a 10-year, random stock study against the market random walk, regression, auto-regression, and multiple neural network models typically used in prior studies. By combining human trader techniques with biologically inspired neural network models, Echo ARTMAP may represent a new tool with which to assist in financial decision-making and to explore life-like context sensitivity.