An algo takes a parent-order and produces a stream of child-orders according to a schedule or a set of rules. The trader chooses the objective — track a benchmark, minimise impact, finish by a deadline — and the algo handles the mechanics.
Every algo trades the same three things against each other: market impact, timing risk and the chance of not finishing. Trade fast and you pay market-impact; trade slow and the price may move away before you are done.
Example: 500,000 shares in a stock that trades 4 million a day. A vwap-algo spreads it over the session near 12% of volume; a sniper-algo waits for liquidity and may finish in twenty minutes or not at all. Both beat one 500,000-share market-order.
Related: parent-order, child-order, twap-algo, vwap-algo, implementation-shortfall