A financial trading engine can be fast and still be dangerously wrong. That was one of the key engineering lessons from building a low-latency financial order-matching engine in Rust. In financial systems, performance matters. But incorrect order execution, broken price priority, or inconsistent state can create risks that speed alone cannot solve. I designed the engine around a single-writer architecture, bounded message passing, integer-based pricing, FIFO order matching, and self-trade prevention. The implementation also explored cache-efficient memory layouts, lock contention, and predictable execution latency. The results from local synthetic benchmarks: • p50 latency: 48 nanoseconds • p99 latency: 56 nanoseconds • p99.9 latency: 69 nanoseconds • All 14 integration tests passed. These measurements reflect isolated engine operations, not end-to-end trading performance. One important lesson stood out: Low latency is a performance characteristic. Financial correctness is a system requirement. The next challenge is ensuring deterministic recovery, durable execution records, and reliable risk controls. How do you balance performance and correctness when designing financial systems?