Prix, Distinctions & Contrats de recherche
In October 2009, the project "Algorithmic Trading" has received the approval and funding of the French National Research Agency, ANR, under the reference ANR-09-JCJC-0139-01. It is a four-year "Jeunes Chercheurs - Jeunes Chercheuses" project coordinated by a young researcher, Sophie Moinas.
The project was launched on October 1, 2009 for an initial period of 4 years, and it has benefited from a time extension of 6 months to be closed on March 31, 2014 (that is, a total period of 54 months). It has received a financial support of €100,000, and an additional support of €30,000 dedicated to the potential compensation of reductions in teaching loads. Expenses total €91,606, and the members of the team have not benefited from reductions in teaching loads.
The initial objectives of the project
Stock market activity used to be concentrated on floors. Investors would use the telephone or telegraph to send their orders to traders present on the floor. These traders would then execute the orders via face to face interaction with other market participants. Progress in communication technologies has changed this. Most exchanges are now electronic. Investors and traders based far from the exchange can rapidly observe quotes and trades on their screens. And they can feed orders to the market via their computers. In recent years, financial institutions have indeed developed programs to automate most of their trading strategies. Algorithmic trading may be defined as the automated, computer-based execution of orders via direct market-access channels, usually with the goal of meeting a particular benchmark. It should be distinguished from ‘program trading’. The later is the computer-driven, automatic execution of buy and sell orders for a basket of securities (typically, more than 15 stocks), for instance when the potential loss exceeds a predefined threshold. In contrast, algorithmic trading refers to a form of automated trading in which computers execute orders on individual stocks, based on a series of parameters (e.g. time, price, volume).
The phenomenon of algorithmic trading is growing fast. According to Tower Group, 27% of hedge funds and 16% of Institutional Investors were using algorithmic tools in 2006.1 While mostly U.S. equity markets were concerned to start with, algo trading progressively expands to European markets. In Europe, it represents 45% of the trades in the electronic limit order book Xetra in the first quarter 2008, while it represented only 39% of trades in 2007. Algo trading is also developing in more complex derivatives, futures and option markets, and even in bond markets.
While financial innovation contributes to increasing welfare, for instance by matching products or services to the needs of either firms, investors, or financial institutions, it often yields to financial dysfunctionnalities. For instance, ‘securitisation’ is currently said to have widely participated to the subprime phenomenon, and the subsequent financial and banking crisis. Algorithmic trading is a huge innovation for the financial sector. One may therefore wonder what could be the potential improvements and risks linked its development. The massive use of algorithmic trading in financial markets raises important questions. The main question seems to be the following: how does algorithmic trading affect markets? In particular, does it increase or reduce liquidity? Does it improve the informational efficiency of prices? Who benefits from algorithmic trading? Does it alleviate adverse selection problems or does it make them more severe? The answer of this main question may have policy implications for the regulators.