Paper: arXiv 2611.08701

Authors: Hieu Tran, Hong Yu

Abstract

Execution algorithms are calibrated asset by asset, because the queues, trade rates and bid-ask spreads they face differ by orders of magnitude across assets. For reinforcement learning this is zero-shot generalisation across contexts of very different scales, and microstructure invariance proposes a normalisation fixed by theory: measured in each asset’s own units of time, size and price, markets look alike. In an exact order-by-order replay of public Nasdaq data, we train policies once, on 954 stocks in 2019, to choose between waiting in the queue and crossing the spread, and evaluate them unchanged on 1,964 unseen stocks in 2020-21, 2,629 stocks in 2025-26, crypto pairs and a second exchange. First, a policy that sees a stock only through its recent trade rate, trade size, spread and price increment (tick) costs within 0.015 spreads per share (a fifth of what learning saves) of a policy whose training data included the stock, and the stock’s own history does not help it; it comes within a similar margin of policies trained on the second exchange when every episode counts equally, and within a wider one on crypto. Second, in mean cost before fees, invariant units beat raw units, z-scores and asset descriptors, but by under a basis point: most of their advantage is information and a common cost scale that a learner can also take as plain inputs, and none of the representations tested extrapolates to unseen spread regimes. Third, a near-zero transfer gap to a pooled comparator is weak evidence that a representation is right: we show that such a comparator detects only a small share of what a poor representation loses, and confirm it in synthetic markets and on pendulums.

Complexity vs Empirical Score

  • Math Complexity: 6.5/10
  • Empirical Rigor: 9.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: This paper demonstrates strong empirical rigor through extensive backtesting on real Nasdaq data and multiple asset classes. Its novelty lies in applying microstructure invariance to zero-shot reinforcement learning for execution policies, a significant step towards generalizable trading algorithms. The methodology is well-detailed, supporting high reproducibility.

Research Flowchart

  flowchart TD
    A[Research Goal: Zero-Shot Transfer of Execution Policies Across Diverse Assets?] --> B{Methodology: Microstructure Invariance & RL}
    B --> C[Data/Inputs: Nasdaq Public Data (2019), Unseen Stocks (2020-21, 2025-26), Crypto Pairs, Second Exchange]
    C --> D[Computational Process: Exact Order-by-Order Replay, Train Policies Once (954 stocks), Evaluate Unchanged (1,964+ unseen)]
    D --> E{Finding 1: Invariant Units Enable Near-Zero Transfer Gap}
    E --> F[Finding 2: Invariant Units Beat Raw Units/Z-scores by <1 BPS (Information Advantage)]
    F --> G[Finding 3: Zero-Transfer Gap to Pooled Comparator is Weak Evidence of Representation Quality]