Date of Award
Spring 2026
Degree Type
Honors College Thesis
Academic Program
Computer Science BS
Department
Computing
First Advisor
Dr. Zhaoxian Zhou
Advisor Department
Computing
Abstract
Market prediction attempts have primarily focused on large-cap stocks due to their stability and market consistency. As such, studies that use time-series techniques to predict large-cap stocks have produced consistent results. Despite the success of large-cap predictions, penny stocks have remained unexplored in modern academia due to their high volatility, low liquidity, and structural instability. Regardless, unexplored market potential and technological advancements underscore the need for preliminary research into penny stock forecasting. This study aims to determine whether meaningful predictive structures exist in time-series penny stock data. This study utilizes an incremental approach. Various penny stocks were selected, pooled, and preprocessed. The final feature set was then fed into a logistic regression model to determine if a decision boundary or feature correlation existed. Results obtained from the logistic regression were refined and contextualized using autoregressive frameworks and macroeconomic indicators. The final results do not support the existence of easily extractable structures to accurately and consistently predict penny stock movements under the given experimental design. Likewise, penny stocks exhibit little sensitivity to broader economic indicators and fail to adhere to common statistical assumptions. These findings suggest that structural constraints and internal noise likely limit the accuracy of penny stock forecasting. Moreover, successful penny stock prediction might require the construction of unique features to overcome these barriers.
Copyright
Copyright for this thesis is owned by the author. It may be freely accessed by all users. However, any reuse or reproduction not covered by the exceptions of the Fair Use or Educational Use clauses of U.S. Copyright Law or without permission of the copyright holder may be a violation of federal law. Contact the administrator if you have additional questions.
Recommended Citation
Hait, Susom, "Evaluating Predictive Structure in Penny Stocks Using Machine Learning and Statistical Methods" (2026). Honors Theses. 1093.
https://aquila.usm.edu/honors_theses/1093