I'd sell you my kids before I'd sell you my data, and I'm not selling you my kids.
—Anonymous quantitative futures trader
The old adage is “garbage in, garbage out,” meaning that if you use bad inputs, you'll get bad outputs. This relates to quant trading because most quants utilize some form of input/output model, which is a term that comes from computer science (and that has been borrowed by econometricians). It refers to the way in which information processors (such as computers) communicate with the world around them. One of the things we love about input/output models is that if you provide the same input a million times, the output should be consistent every time. The process that transforms an input into an output is typically the part that people call the black box in quant trading, and we have seen the inside of this box in the preceding chapters. In this chapter, we examine the inputs of quant trading models, namely, the data they depend on.
Mechanically, data reach the black box through data servers, which are connected to one or more data sources. On receipt of these data, the black box processes them for use by the alpha, risk, transaction cost, portfolio construction, and execution models that constitute the internal organs of the quant trading machine. These data servers usually process data using software some quants call data feed handlers, which are designed to convert the data to a form in which they can be stored and utilized by the modules of ...