You need to think more in terms of descriptive statistics first and why you would approach the problem in a particular way. If you are data scientist that throws in everything and looks for what would stick then you are in the wrong place. AQR is the kind of place where economics and prior is more (way more) important than any other place.