Validation of MDQR on Small-Tick Assets

Determine whether the Multidimensional Deep Queue-Reactive (MDQR) model accurately reproduces key small-tick market characteristics, including spread distributions and regime transitions, by validating the MDQR framework on small-tick assets and benchmarking its performance against established generative models for such markets.

Background

The paper develops the Multidimensional Deep Queue-Reactive (MDQR) model and demonstrates strong performance on a large-tick, highly liquid futures market (Euro-Bund). In comparing MDQR with other deep learning approaches, the authors note that some alternative models have been shown to handle small-tick assets effectively, especially in reproducing spread distributions and regime transitions.

However, the MDQR framework has not been validated on small-tick markets. Establishing its effectiveness in this setting would clarify the model’s broader applicability and address whether its benefits extend beyond large-tick instruments.