Optimizing queue position in market making strategies The price-time-priority design of most exchange matching engines requires high frequency trading firms to continue evolving techniques for front-of-line positioning. On non-pro-rata order books, incoming orders interact with the oldest liquidity in a price level first and the youngest liquidity last. Orders that are filled at the end of the queue exhibit negative mark-to-market P&L at short time frames, as the new mid-price establishes in place of the removed level. Adverse selection associated with bad queue position leads market makers to ensure orders maintain highest priority while canceling orders with low priority. Queue position optimization techniques are based on study of order protocols, market data packet sequencing, book forensics, and latency optimizations, from specialized order types to ladder placement strategies. Examples: • Good ‘Til Canceled (GTC) orders: Exchanges preserve GTC orders in memory during closing hours and reinstantiate the orders automatically when the exchange opens again. Some firms maintain deep orders for weeks or longer to ensure they’re first in the queue when the market price drops or rises to the market maker’s level. • Hidden order types: A handful of US equity exchanges support specialized order types such as “Hide Not Slide” and “Hide And Light.” These keep orders hidden and non-executable while crossing the BBO, then immediately unhide them at the front of the queue once the book uncrosses. • Filling in empty levels: If an order book’s BBO is more than one tick wide or has gaps between prices, market makers will fill in empty levels purely to secure queue position. Trading strategies will often disregard side or price with respect to fair value when presented with the opportunity to establish a new price level. • L3 feed parsing: All HFT market making systems subscribe to order-by-order (level 3) market data feeds, the most granular data that exchanges provide. The data stream allows the consumer to create the exact book of orders the matching engine sees, and enables the trader to understand their exact position in the book after the order has been acknowledged. • Size fingerprinting: As a computational shortcut, some firms will send orders at random odd-lot sizes in order to identify the firms’ order locations in public data feeds. One downside of this approach is the possibility that other participants can detect the orders as well and copy the market makers’ strategies. • Racing after fills: One of the most competitive speed games in trading is racing after a large fill that takes through one or more price levels, with the goal of establishing the new best bid or offer in the same direction as the liquidity-taking order. On traditional colocated exchanges, FPGAs are a minimum requirement to reach the nanosecond-scale latency to have a chance at this strategy. Several venues have attempted to use periodic batch auctions and pro-rata systems to incentivize market participants away from speed competition and order type jockeying. Those market mechanisms have largely failed because they’ve led to empirically worse price discovery for institutions and retail brokerages. As agentic trading systems enter markets, efficient, time-tested, predictable market design will matter even more.