Emergence of Randomness in Temporally Aggregated Financial Tick Sequences

Battle Count: 68

The paper is relevant to quantitative trading primarily through its assessment of market efficiency and predictability at different time scales. Understanding where randomness emerges (or fails to emerge) in tick data informs execution algorithms, microstructure-aware strategies, and the detection of algorithmic trading patterns. The finding that high-frequency stocks retain predictability even at high aggregation levels could inform HFT strategy design. However, the paper does not propose specific trading signals or strategies.

The methodology is conceptually straightforward (binary encoding + standard test batteries) but requires careful parameter selection, handling of string length constraints, and interpretation of multiple test results across categories. The sanity check procedure adds complexity. Implementation requires access to LOBSTER data, NIST STS, TestU01, and custom entropy test code. The online nature of the encoding simplifies real-time application.

GitHub repository provided with full code and results. Standard test batteries (NIST STS v2.1.2, TestU01 v1.2.3) are publicly available. Parameters are fully documented in tables. However, the LOBSTER data source is commercial and not freely available. The methodology is clearly described with step-by-step algorithms for entropy tests and binary encoding.