Alpha in the Analytics: Optimizing Algorithmic Strategies
Yesterday’s demand for a cockpit of controls is shifting to the need for better auto-pilot functions with monitoring tools and the ability to change course in the case of turbulence. Does this sound like a high-touch service? It is, but for the low-touch. The goal of algorithmic optimization is to deliver the kind of symbiotic relationship for low-touch trading that has always existed for the high-touch side. There is an opportunity to deliver the low-touch version of their top sales traders – a well-rounded picture of the market, approaches on how to trade the order, a constant feedback loop and the ability to change course at any time, based on their intent.
Algorithmic trading in today’s fragmented market place poses many challenges for a buy-side trader simultaneously managing multiple orders. The decision about which algorithm to use demands a full understanding of the algorithms available and a constant need to ensure that at every moment during the order’s lifecycle, the optimal algorithm or algorithmic strategy is being deployed.
Tactical algorithms such as VWAP or TWAP are broadly available – workhorse algorithms that have a high level of predictability but limited purpose and function. Constantly optimized, they are reactive execution tools that accomplish a number of objectives, such as trading a basket of stocks, a futures hedge, or in conjunction with other execution strategies such as trading a block in the dark. For the most part, there may be little difference in the outcome from one broker’s algorithm to the next and the onus is on the trader to constantly switch in and out of such algorithms as market conditions and the state of the order change.
Straddling individual broker algorithms is the notion of algorithmic optimization, using an engine which uses predictive analytics to switch automatically between algorithms. This enables an algorithmic strategy or intent to be set, and for the optimization engine to use all the algorithms available within a given broker’s suite for a particular order. In essence, this mimics the trader manually switching among a set of algorithms.