A liquidity motivated algorithm for discerning trade direction

Most exchanges do not report trade direction thus researchers and traders must deduce whether a trade is buyer or seller initiated since this information is required to evaluate models of bid-ask spread components and to understand the market for immediacy. Algorithms that assign trade direction based on the proximity to bid or ask quotes are easily implemented but ignore information readily discernable from orders, changes in the quoted depth and subsequent price movements. Using the New York Stock Exchange Trades, Orders and Quotes database, systematic biases in existing trade direction algorithms are documented that can be rectified by recognizing that the impact on liquidity is the fundamental characteristic underlying order placement. Although this liquidity-based method is difficult to implement, it more closely captures the actual behavior of market participants.