Completed an open ended data task based on their dataset, which contained much of the data they extract and attach with a sentiment and other features. It was poorly defined time periods across which returns were calculated not explained, labels of assets or sentiment extraction were not explained. Wrote 3 iterations of 2 different models. Took about 10 hours cumulatively, generated alpha. These guys replied with really minimal feedback would refuse to follow up my questions. Even though I provided them a working and successful model prediction with a framework. Seemed to care more about my writeup and the fact I used Python. They claim to be used by 7 top quant funds - yet it took them for 2003-2006 3 years to come up with their API. I have designed a 6 situation (corporate) detector in under 6 months by myself. I don't understand what took them so long. Very bad experience I think they're fundamentally doing a sub optimal job and like many startups use this Data Task as an information mining exercise. Don't be fooled, also look up the profiles of the founders not such high fliers.