Date of Award
5-2025
Degree Type
Doctoral Dissertation
Degree Name
Doctor of Business Administration (DBA)
Department
Jack Welch College of Business & Technology
Dissertation Supervisor
Dr. Loran Chollete
Committee Member
Dr. Mahfuja Malik
Committee Member
Dr. Md. M. Hasan
Abstract
The purpose of this paper is to provide traders with a trader friendly model that would enable them to accurately predict Bitcoin’s high price and low price so that they are able to make more informed decisions for improved risk management when trading the highly volatile asset – Bitcoin. To achieve this purpose, this paper poses the following research question: Which statistical model-frequency combination best predicts – in terms of Mean Absolute Percent Error (MAPE), Akaike Information Criterion (AIC), and Schwart Information Criterion (SIC) – Bitcoin’s high price and low price? This paper also poses the objective of ensuring that the research question is both answered and that the statistical model-frequency combination selected is trader friendly or in other words, both easy to implement and interpret. To answer the research question and to meet the objectives of this paper, this paper develops two types of statistical models – Ordinary Least Squares (OLS) and Autoregressive Distributed Lags (ARDL). These models are developed across multiple frequencies – daily, weekly, and monthly. These statistical models and frequencies are chosen due to their trader friendliness, which aligns with the objective of this paper. Results for each of these models training sets show that the ARDL Bull Weekly model best predicts Bitcoin’s high price while the ARDL Bull Monthly model best predicts Bitcoin’s low price. To test the robustness of these two models, each model is tested on 7 robustness tests that incorporate out of sample data that covers various market conditions from 2011 to 2024. Results for the robustness tests show that both the ARDL Bull Weekly model and ARDL Bull Monthly model are highly robust for the out of sample data and market conditions they are tested on. This is evident in both models’ remarkable average mean absolute percentage error (MAPE) across out of sample robustness tests of 1.3% (ARDL Bull Weekly) and 3.28% (ARDL Bull Monthly). This indicates that these models could be used by traders to accurately predict Bitcoin’s high price and low price so that they are able to make more informed decisions for improved risk management.
JEL Classification
C01, C02, C52, C53, C58
Recommended Citation
Rice, T. (2025). Optimal prediction of bitcoin high and low prices: An exploratory analysis. Jack Welch College of Business & Technology dissertation, Sacred Heart University, Fairfield CT. Retrieved from https://digitalcommons.sacredheart.edu/wcob_theses/48/
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Included in
Business Administration, Management, and Operations Commons, Corporate Finance Commons, Finance and Financial Management Commons
Comments
Submitted as partial fulfillment of the requirements for the degree of Doctor of Business Administration in Finance Sacred Heart University, Jack Welch College of Business and Technology, Sacred Heart University
DISSERTATION Number DBA05/2025