Project CryptOL
GitHub Link
Project CryptOL™ code and documentation are available on GitHub.
This webpage provides a comprehensive description of the ongoing development work on CryptOL™, aiming to predict cryptocurrency prices using advanced AI algorithms. Stay tuned for further updates as the project evolves towards a full alpha release.
Abstract from the Original Paper
Title: Using Artificial Intelligence to Predict Cryptocurrency Price with Project CryptOL™ Software Application
Authors: Alexander Gribtsov, Chris Woszczak, John Guseman
Abstract:
Background: Implementation of different types of artificial intelligence algorithms can produce fit data that predicts the price of different cryptocurrency types such as Bitcoin.
Related Work: Two related projects were observed for cross-comparison of data performance and overall cryptocurrency price forecasting experience.
Architecture: Project CryptOL is a web-based full stack application. The design, implementation, and deployment of this project are covered in detail.
Testing: Embedded into the CryptOL application is a classical false negative and false positive testing technique of fitness of running data.
Discussion: During the development of the project, all team members became familiar with common data science terminology and standardized techniques.
Conclusions: Project CryptOL is a great example of how standardized data science techniques and algorithms can be applied together to present a strong indicator of future cryptocurrency price.
Keywords: Algorithms, artificial intelligence, bitcoin, cryptocurrency, machine learning, online machine learning, LSTM, neural networks, data science, survival modeling, time series, linear regression, logistic regression.
Original White Paper
Prediction Algorithm in Action: image.
Project History
2020: Initial development started by Alexander Gribtsov, Chris Woszczak, and John Guseman.
2020: Project was abandoned.
2023: Development resumed with a renewed focus on integrating multiple AI models for better prediction accuracy.
Project Components
Architecture:
CryptOL™ is a web-based full stack application.
It uses a singular design to tie the components together.
Data is obtained using the Yahoo Finance API.
Algorithms:
Linear Regression: Predicts the value of Bitcoin based on selected features.
Time Series (ARIMA): Forecasts future values based on known past values.
Long Short-Term Memory (LSTM): A form of recurrent neural networks for long-term predictions.
Data Generation and Splitting:
Collects historical Bitcoin prices based on the prediction interval.
Splits data into training (80%) and test sets (20%).
User Interaction:
Users interact through a GUI built using Anvil.
Predictions are made based on selected algorithms and time intervals.
Testing:
Evaluates models based on directional accuracy of predictions.
Uses false negative and false positive testing techniques.
Deployment:
Hosted on Anvil for reliability.
Updates based on user feedback occur biweekly.
Timeline
2020: Project inception and initial development.
2023: Development resumed.
2024: Current pre-alpha state with public interaction available.
Roadmap
2025 Q1: Enhance GUI and improve user experience.
2025 Q2: Integrate additional AI models for cross-comparison.
2025 Q3: Conduct extensive testing and validation of predictions.
2025 Q4: Full alpha release with a comprehensive set of features.
Login Window: image.
Algorithm Description: image.
Screenshots
General UI: image.
Project CryptOL™: Cryptocurrency Prediction Using Artificial Intelligence
Overview
Project CryptOL™ aims to predict cryptocurrency prices using artificial intelligence algorithms. Initially developed in 2020 by three members of the Computer Science Department at Monmouth University, the project was abandoned and resumed development in 2023. The project is currently in its pre-alpha state, and public interaction is open here.