SOR Venue Selection Logic

How to Read This Diagram

The SOR engine processes each incoming order through a multi-stage evaluation pipeline. First, it collects live quotes from all connected venues (ECNs, LPs, internal books). Then it normalizes these quotes by adjusting for trading fees, expected latency to each venue, and credit availability.

The Scoring stage applies a weighted multi-factor model that considers price, fill probability (based on historical data), expected slippage, and venue reliability. The venue with the best composite score is selected, and the order is routed.

Typical Scoring Factors

• Price (40-60% weight): Raw bid/ask after fee adjustment.
• Fill probability (15-25%): Historical fill rate at this venue for similar order sizes.
• Expected latency (10-15%): Round-trip time to venue, including expected processing time.
• Slippage history (5-15%): Average slippage on fills from this venue.
• Depth (5-10%): Available liquidity at the quoted price level.

Fallback Logic

If the selected venue rejects the order (e.g., last-look rejection, insufficient liquidity), the SOR does not simply fail. It cascades to the next-best venue in the ranking. Sophisticated SOR implementations maintain the ranking in memory and can reroute within milliseconds. The number of retry attempts and the staleness threshold for quotes are configurable parameters.