TY - JOUR T1 - The Performance Implications of Improving Market Forecasts JF - Trading SP - 40 LP - 48 VL - 2007 IS - 1 AU - Merrell Hora Y1 - 2007/03/20 UR - http://guides.pm-research.com/content/2007/1/40.abstract N2 - This article demonstrates that views about the commoditization of algorithms and the irrelevance of market forecasts are incorrect. The role of forecasts within algorithms is examined, indicating how they directly affect performance. Specific characteristics of high frequency market data that are problematic for conventional approaches are identified, indicating where improvements can be made. Using volume forecasting and VWAP algorithms as an example, it is shown that improving the quality of the forecast significantly improves performance and reduces risk. We demonstrate consistent reductions in average forecasting error ranging from 20% to 33%. These result directly translate into gains in VWAP performance between 19% and 25% with reductions risk of nearly 41%. Since VWAP algorithms are relatively simple in their use of forecasts, it is clear that the link between forecasting accuracy and execution performance generalizes to other classes of algorithms. The implication for algorithm users is that algorithms are not commodities, and that a primary differentiating factor is the quality of the underlying forecasting technologies. ER -