According to Adam Sussman, senior research analyst at TABB Group and the author of a new research note, âThe Modular Algorithm: The Growing Choice in Buy-side Execution Strategies,â the buy-sideâs use of canned algorithms, pegged at 58% in 2005, will account for less than half of all algo flow by 2008 because âthe current usage of various algorithms and DMA tools is not the ideal state for the buy side, but rather the optimal selection based on its current options.â
For algorithm providers, customization further feeds the product pipeline for canned algorithms, re-engages buy-side clients on a personal level â something that had been lost as FIX replaced the phone and entertainment rules became more stringent â and differentiates brands on a crowded shelf of products.
âIn the end, this is a buy-side issue because what traders really want,â Sussman adds, âis a tool that allows them to control an order using a wide range of variables, a tool that permits them to map the portfolio managerâs objectives into an executive strategy. This means an algorithm with a true sense of purpose. Instead of merely wanting all the data they look at incorporated into an algorithm, they want the algorithm to look and react to the data in a specific way.â
Another result of the growing choices among buy-side execution strategies is a steep decline in the amount of resources hedge funds are willing to spend building execution-only algorithms. âThe bare minimum today to build an algorithmic infrastructure,â says Sussman, âis $1.3 million with recurring annual costs of $900,000 and personnel costs account for nearly half of that cost.â In 2005, he adds, there was little choice for a hedge fund to build its own algorithms but with customization services and modular technologies, TABB Group believes that proprietary algorithmic usage will decline from 88% of all hedge fund algorithmic flow to 67% by 2008.
Citing examples, he explains that several quantitative hedge funds chose to split scheduling and routing responsibilities with the scheduling component kept in-house because deciding when to route is dependent upon proprietary data such as the expected alpha among a list of stocks. He adds, however, that these funds are outsourcing routing decisions to their brokers. Using this example, says Sussman, âThe latest generation of algorithms automates this functionality through the use of a âwouldâ option that trades according to a participation strategy but would simultaneously hunt for additional liquidity among dark pools.â
The 20-page research paper covers recent progress in the equity algorithmic space, trends in algorithmic functional areas, hedge fundsâ algorithmic flow distribution, the role of customization within this process, e.g., Algorithm Development Kits (ADKs), the beta release of FIXatdl (FIX algorithmic trading definition language) and Complex Event Processors (CEPs) and future product enhancements based on the custom work being done today. The information for this research report was collected from conversations conducted with buy-side traders and electronic trading executives at sell-side firms in July 2007 and supported by analysis of primary data drawn from existing TABB Group research.