Python for Finance: Analyze Big Financial Data

Further Reading

The paper cited at the beginning of the chapter as well as in the “Conclusions” section is a

good read, and a good starting point to think about hardware architecture for financial

analytics:

Appuswamy, Raja et al. (2013): “Nobody Ever Got Fired for Buying a Cluster.”

Microsoft Research, Cambridge, England,

http://research.microsoft.com/apps/pubs/default.aspx?id=179615.

As usual, the Web provides many valuable resources with regard to the topics covered in

this chapter:

For serialization of Python objects with pickle, refer to the documentation:

http://docs.python.org/2/library/pickle.html.

An overview of the I/O capabilities of NumPy is provided on the SciPy website:

http://docs.scipy.org/doc/numpy/reference/routines.io.html.

For I/O with pandas see the respective section in the online documentation:

http://pandas.pydata.org/pandas-docs/stable/io.html.

The PyTables home page provides both tutorials and detailed documentation:

http://www.pytables.org.

[ 28 ]

Here, we do not distinguish between different levels of RAM and processor caches. The optimal use of current

memory architectures is a topic in itself.

[ 29 ]

Another first-class citizen in the database world is MySQL, with which Python also integrates very well. While many

web projects are implemented on the basis of the so-called LAMP stack, which generally stands for Linux, Apache Web

server, MySQL, PHP, there are also a large number of stacks where Python replaces PHP for the P in the stack. For an

overview of available database connectors, visit https://wiki.python.org/moin/DatabaseInterfaces.

[ 30 ]

Cf. http://docs.scipy.org/doc/numpy/reference/arrays.datetime.html.