in computational finance
From August 2011 until February 2013 I was working as a software freelancer
in the computation finance domain.
The main technical topics were:
- Evaluating papers at that time to gauge trends in the quant trading world for a client
- Extensive timeseries analysis on specific markets with tools like R, Matlab, Python etc.
- Implementing trading algorithms and tools in C++/Kotlin/Python
Since I signed a formal obligation of confidence, I cannot be too specific here.
Overall I can say that I was very lucky to get in touch with a client who
had specific needs for trading tools.
I took the opportunity for a temporary contract and became a freelancer.
The client was so happy with my implementation for the needed tools, that he
asked me to implement also a live trading algorithm.
Amongst others I researched the following topics
- Optimal stopping
- Mean reversion
- ML and AI
- Data collection and preparation
Implementing a trading algorithm connects many aspects of software design.
Not only is the logic itself important, but special awareness has to be put on
- High-throughput on financial data
- Model scalability to incorporate other markets and opportunities on the fly
- Plugin architecture design for hot-swapping models
- High-availability which means handling of errors in a graceful way
For evaluating ideas and present them visually, I mainly used R which has
excellent stastical and plotting support.
When it comes to implement the ideas, I used the popular libraries in Python
like NumPy, Pandas, SciPy etc.
The final products were then implemented in C++ and Kotlin which make use
of the forementioned libraries at runtime.