The Algo Wheel

Before algo wheels, a buy-side trader routing an order to a broker's execution algorithm made that choice based on relationships, habit, or gut feel about which broker "does well" in a given name — a process that's hard to defend objectively and nearly impossible to audit for best execution. An algo wheel replaces that discretion with a systematic rotation: orders meeting defined criteria (say, a certain size and liquidity bucket) get automatically assigned to one of several pre-approved brokers' algorithms in a randomized or weighted rotation, removing the trader's case-by-case judgment from routine flow.

Because every broker receives a comparable, randomized sample of similar orders rather than a self-selected subset, the resulting execution quality — measured through transaction cost analysis (TCA) against benchmarks like arrival price or VWAP — becomes a fair, apples-to-apples comparison between brokers. Brokers that consistently underperform on the wheel get their allocation weight reduced or removed; brokers that perform well get more flow, turning execution quality into a continuously monitored, evidence-driven feedback loop instead of a relationship-based decision.

Traders still handle the flow that doesn't fit the wheel's criteria — large blocks, illiquid names, anything needing manual judgment — so the wheel typically covers a defined slice of routine, algorithm-appropriate orders rather than every trade a desk sends out.