Limit order book simulators need to reproduce the fine-grained timing of real markets to be useful. Getting the price dynamics right is necessary but not sufficient — if the simulator's temporal structure doesn't match reality, execution strategies that work in simulation will fail in production. The timing isn't decoration on top of the price process. It's part of the mechanism.
Noble, Rosenbaum, and Souilmi (arXiv:2603.24137) build a simulator for large-tick assets that compresses book state into spread and volume imbalance, then calibrates temporal dynamics to real data. The striking feature is a pronounced mode in the inter-event time distribution at the exchange's round-trip latency — the time it takes a message to travel to the exchange and back.
This mode reveals coordinated behavior. When news arrives, many market participants react simultaneously. Their responses travel to the exchange, arrive within a narrow time window determined by the round-trip latency, and create a burst of activity. The mode is not random clustering — it's the fingerprint of parallel reaction to shared information, filtered through the exchange's physical infrastructure. The speed of light and the length of the cable create a temporal pattern as characteristic as a spectral line.
The simulator reproduces both concave market impact during execution (early shares move the price more than later ones) and partial post-trade reversal (the price recovers some of the impact after the trade completes). These features emerge from a directional accumulation mechanism with exponential decay — a simple model that captures the difference between temporary and permanent impact.
The practical test is whether simulated strategy returns match real ones. They do, with strategy PnL depending critically on execution choices — the simulator is sensitive enough that different execution algorithms produce meaningfully different outcomes, which means the simulation is capturing the features that matter for real trading decisions.
The latency mode is the most structurally interesting finding. Market microstructure is usually described in terms of information, incentives, and strategy. The round-trip mode shows that physical infrastructure — the speed of signal propagation, the location of servers — imprints itself directly on the statistical structure of the order flow. The physics of the network is visible in the statistics of the market.