CME Fills: MDP 3.0 Trade Summary vs iLink private Fills latency¶

It is commonly understood that CME private fill confirmations are delivered ahead of public market data. In this study, we set out to validate and quantify this assumption using real measurements.

By matching market data events with corresponding private confirmations, we observed several noteworthy results.

• Convenience Gateway (CGW) median latency advantage: 16 microseconds
• In approximately 89% of cases, CGW confirmations arrived before Trade Summary

• However, in 11% of cases, CGW confirmations were delivered after public Trade Summary data

• Market Segment Gateway (MSGW) median latency advantage: 64 microseconds

Overview¶

This study compares latency across three CME data paths:

• Trade Summary (Market Data) - public MDP3 feed
• MSGW (Market Segment Gateway) - private direct exchange confirmation (ExecutionReportTradeOutright iLink message)
• CGW (Convenience Gateway) - private routed confirmation (ExecutionReportTradeOutright iLink message)

The goal is to validate expected latency relationships and quantify real-world behavior.

Expectations¶

Before running the analysis, the assumptions were:

1. MSGW should be the fastest (direct segment gateway)
2. CGW should be ~30µs slower than MSGW (based on CME Public Documents )
3. Both private confirmations (MSGW, CGW) should be faster than Market Data

Methodology¶

We matched Trade Summary events from MDP 3.0 feed with private confirmations based on Transaction Time and Order IDs. For each such event we calculated CME latency for both Market Data and iLink.

Latency is measured using CME-provided timestamps:

Latency = Exchange Sending Time - Transaction Time

More at CME Globex Timestamps

This ensures consistency across all message types and avoids local clock bias and exchange side matching latency variability which will affect both Market Data and iLink messages.

Results¶

MSGW vs Market Data¶

Key Takeaways¶

• MSGW is consistently faster than Market Data
• Median improvement: ~64µs
• This confirms expected behavior:
• Private confirmations arrive before public dissemination

CGW vs Market Data¶

Key Takeaways¶

• Median benefit: ~16µs
• CGW is only marginally faster than Market Data
• High variance observed

Unexpected Finding¶

CGW Slower Than Market Data¶

Percent of CGW slower than Trade Summary: 11.04% Events: 56 / 500

This contradicts the expectation that private confirmations always lead public data.

Summary¶

MSGW Behavior¶

• Predictable and consistent
• Clear latency advantage (~64µs median)
• Should be used when possible for latency-sensitive strategies

CGW Behavior¶

• Higher variance and routing overhead
• Sometimes lags behind Market Data
• More modest median advantage (~16µs) with high variance.

Why CGW Can Be Slower¶

Possible contributing factors:

• Internal routing and aggregation within CGW
• Shared infrastructure and queuing
• Additional processing layers compared to MSGW
• Market data path occasionally benefiting from parallelization or prioritization

Practical Implications¶

Strategy Sensitivity¶

• Ultra-low latency strategies
• Should rely on MSGW

• CGW may introduce unpredictable delays

• Moderate latency strategies

• CGW may still be acceptable
• But assumptions about always beating market data are incorrect

Conclusion¶

• MSGW behaves as expected and consistently leads market data
• CGW does not guarantee latency advantage
• In ~11% of cases, market data arrives before CGW fills

This highlights an important takeaway:

Not all "private" paths are equal - direct exchange access (MSGW) is fundamentally different from routed access (CGW).

Final Thought¶

For latency-critical trading:

• MSGW = deterministic edge
• CGW = convenience with trade-offs
• Market Data = surprisingly competitive with CGW in a some scenarios