TradeZen Emotion-reactive Interface for Financial Traders

We propose a wearable sensor that constantly monitors various physiological metrics to gauge emotions and stress, then broadcast emotion-events to devices and applications that would proactively tailor their behavior accordingly. Specifically, we attempt to build a wearable stress-sensor that broadcasts stress-events wirelessly to trading software, either portable or on the desktop, so that the software requests explicit rationale for trades seemingly executed under stress. Our contributions are: A) Identifying that trading and other stress-sensitive activities do not need to have a wearable interface, but only the subconscious, first-person act of sensing moments of stress requires wearability, and B) Proposing that interfaces for stress-sensitive activities should pro-actively tailor itself to the stress level of its users, and C) Devising a scheme for a wearable stress sensor to update stress-sensitive applications wirelessly. We construct a low-fidelity prototype arm-band with Galvanic Skin Response sensors, and a bluetooth interface that broadcasts stress-events to an application on an Android phone. The application is a prisoner’s dilemma game, instead of the proposed trading software, to facilitate repeatable experiments. We evaluate the device’s effects on the consistency of decisions made under stress and conflicting evidence via observation of 12 subjects playing 3-player iterative PD games. The pay-off matrix changes to favor different strategies after random number of rounds, and we gauge the degree of consistency of the players’ move patterns. Our results suggested that the subjects who wore the device and received stress warnings were more consistent in their moves, if perhaps not optimal. However, there were not enough subjects to produce conclusive results. We also found during prototype building that melody and rhythm shifts in music produces very clear GSR responses, which could suggest other applications for wireless emotion event broadcasts.