Symbol ranking compression algorithms are known to achieve a very good compression ratio. Off-line symbol ranking algorithms (e.g., bzip, szip) are currently the state of the art for lossless data compression because of their excellent compression/time trade-off. Some on-line symbol ranking algorithms have been proposed in the past. They compress well but their slowness make them impractical. In this paper we design some fast on-line symbol ranking algorithms by fine tuning two data structures (skip lists and ternary trees) which are well known for their simplicity and efficiency.