Using the human brain as CPU, a new frontier?

When Human computation abilities are applied to automated trading

Arnaud Vincent is a doctor of MINES ParisTech since the 14th of November 2013

Arnaud Vincent’s thesis

Algorithmic trading is about technologies for automatically executing trades on financial markets. Trading robots can now detect market opportunities to sell or buy shares and securities. Machines are replacing humans when it comes to investment decisions. 

However in other fields, researchers working on complex problems such as predicting the 3D shape of proteins given their amino acid sequence, have found that “human computation” outperforms computers. Teams of on-line volunteers with no special knowledge who were invited to play a collective game came up with better solutions to the puzzles than state of the art computer programs. 

The aim of this thesis was to apply “human computation” to algorithmic trading to incorporate input from a crowd of players in the algorithms. Two ideas were explored in this thesis; firstly, data from Twitter was incorporated into trading strategies in the form of alerts, then an open trading game called Krabott was set up to capitalize on the intuition and workpower of hundreds of players.

This work was carried out in Centre of Industrial Economics and Finance at MINES ParisTech under Margaret Armstrong’s supervision.


Dealing with unknown events with Twitter

The first experiment in this thesis is based on the idea that information flows in social media may provide input to algorithmic trading systems based on Human Computing principles. Twitter, the micro blogging platform, was chosen in order to track (1) words that may have an impact of financial markets and (2) unexpected events such as the eruption of the Icelandic volcano. The thesis demonstrates that a significant increase in P&L can be achieved in the second case by treating the unexpected events as alerts, but not in the first.                                      

Krabott a trading self-learning system, based on human brains

The online game Krabott succeeded in getting a community of internet users to optimize parameters of the trading strategies. In this game the “Krabotts” are solutions presented as friendly virtual bots each containing a specific set of parameters for a particular trading strategy in their DNA. Humans playing the game, interact in the selection and reproduction steps for each new “Krabott”.

 Research results

Results on live-testing on the FOREX market (currencies) show that “human computation” outperforms machine calculations in optimizing the real-time parameters of trading strategies and providing valuable information on unexpected events. This suggests that it may be possible to set up a hybrid human-computer system based on Human Computation where each player is a kind of single CPU within a global trading system and reach new levels of complexity and efficiency.

The thesis also provides new methodologies and tools for researchers in dynamic and intelligent datamining that may be applied in other areas (medicine, biology, climate, etc…).  

A thesis after 15 years of professional experience… why ?

Arnaud Vincent passed his doctorate on November 14th 2013. His thesis was focused on “Human computation applied to algorithmic trading”.

Arnaud obtained his master’s (DEA) at the Ecole Normale Supérieure de Cachan, in automated systems for manufacturing in 1998.  Over the past 15 years he worked in the racing tyre division of MICHELIN and as the CEO of an IT start-up in the early 2000’s (in datamining) before joining a subsidiary of BPCE group to act as the CEO of an IT company and head of innovation for the past 7 years.

He decided to prepare a PhD thesis in the Cerna Centre of Industrial Economics and Finance at MINES ParisTech to carry out original research while working at the BPCE group. A PhD was the best way for him to further his research during 4 years, outside any short term business or commercial constraints. Now he plans to launch a new IT company in order to provide very specific products based on human computation and consultancy for hedge funds and banks.