Modeling Trade-Size Distributions and Price Impact for Order Placement

Summary

This analysis studies aggregated buy-trade amounts and price movements using exchange trade data. It resamples individual transactions into fixed time intervals and tests whether the distribution of total traded quantity can be approximated by scaling a model fitted to single-trade sizes. The fit is reported as reasonable for short intervals, with greater errors as intervals lengthen; the author attributes this partly to a correction in the Pareto-based formula and discusses a simpler alternative for aggregated trades.

The article also estimates price impact from individual trades and fixed intervals, reporting that most observed trades do not move price and that larger amounts tend to accompany larger moves. It then builds a simplified expected-return model to suggest an order placement size between very small and very large amounts. This is exploratory, based on a limited sample and lacking order-book depth. The model assumes price reversion and simplified order flow, while real order arrivals cluster and market depth changes over time, so its suggested placement is not a validated optimum.

Key ideas

• The distribution of interval-level traded quantity can be approximated from single-trade size behavior, with fit quality depending on interval length.
• Trade amount and observed price impact show an approximate relationship in the sampled data.
• The estimated probability of execution at a given depth is distinct from actual fill probability because the order book changes while an order waits.
• The expected-return model balances trade size, execution probability, and estimated impact to motivate an intermediate order placement.
• The proposed placement model is tentative because it assumes price reversion, simplifies order flow, and omits depth data.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.