Exploring Behavioral Patterns in Small-Cap Stocks and Developing Actionable Investment Strategies

Abstract

This thesis investigates whether short-term behavioural and microstructural inefficiencies in U.S. small-capitalisation equities can be modelled and exploited through systematic intraday trading strategies. Five rule-based strategies were developed, each targeting a distinct pattern commonly observed in retail-driven small-cap movements, including premarket exhaustion, VWAP failures, gap-momentum reversions, and failed breakouts. Historical backtests were conducted using FlashResearch, while cross-validation and real-time testing were performed through a Python-based alert engine connected to Polygon’s live 1-minute feed.

The results show that all strategies exhibit consistent and statistically meaningful behaviour across both historical and real-time datasets, with the underlying logic proving reproducible despite minor feed-specific differences. An exploratory combined analysis suggests that the strategies operate at different points of the trading session and therefore complement one another, while a benchmark comparison with the Russell 2000 highlights their independence from broader market direction.

Although the models demonstrate strong potential, further refinement is required before operational deployment. In particular, full Polygon-based backtesting, correlation analysis, and robustness testing are needed to quantify the exact magnitude of performance. Overall, the thesis provides a practical and academically grounded foundation for the development of a scalable, market-neutral, intraday trading framework.