Q4 2025 NDD Broker Execution Benchmark

All tests were conducted using standardized 0.01 lot market orders on EURUSD, GBPUSD, and USDJPY during London and New York session overlaps (13:00-17:00 UTC). Orders were placed at 30-second intervals from a VPS co-located in London (LD4) and New York (NY4).

Each broker received approximately 3,000 test orders distributed evenly across the three currency pairs and both VPS locations. Orders were placed using the MT5 API with nanosecond-precision timestamps for latency measurement.

Fill ratios ranged from 93.1% to 99.2% across the 15 brokers tested. The top quartile (4 brokers) maintained fill ratios above 98%, consistent with genuine STP/ECN execution. Three brokers showed fill ratios below 95%, warranting further investigation into their liquidity routing practices.

Notably, brokers advertising ECN execution did not consistently outperform STP brokers. The highest fill ratio (99.2%) was achieved by an STP broker, suggesting that the quality of liquidity provider relationships matters more than the execution label.

Median execution latency (P50) ranged from 28ms to 142ms. The wide spread reflects differences in broker infrastructure, bridge technology, and LP response times. VPS co-location in LD4 consistently produced lower latency than NY4 for European-headquartered brokers.

P99 latency (worst-case) showed even greater variance: from 85ms to 680ms. This tail latency is critical for scalpers and high-frequency strategies, as it represents the conditions under peak load or volatile markets.

• Execution labels are unreliable -- Brokers advertising ECN did not consistently outperform STP brokers on any metric.
• LP quality is the key differentiator -- Brokers with Tier-1 bank LP relationships showed 15-20% better fill ratios.
• Tail latency matters -- P99 latency varied by 8x across brokers, far more than P50 latency (5x).
• Co-location benefits are real but modest -- LD4 VPS reduced median latency by ~12ms vs NY4 for European brokers.