A method for performing adaptive lossless data compression using a buffer memory comprises the steps of receiving a plurality of first match signals (m(0) to m(n)), each first match signal indicating whether a byte of an input sequence matches a byte in a memory location of the buffer memory. A plurality of second match signals (mD(0) to mD(n)) are generated, wherein a second match signal (mD) is a delayed version of a corresponding first match signal (m). The second match signals (mD(0) to mD(n)) are used to detect a match sequence between the data bytes received in the input sequence and the data bytes stored in the content addressable memory, and the first match signals (m(0) to m(n)) used to predict whether the match sequence continues during a subsequent byte of the input sequence.