"Algorithm for Predicting Bitcoin Fees" by Mason Tulacz, Matthew Landrum et al.
 

Level of Education of Students Involved

Undergraduate

Faculty Sponsor

Jon Beagley

College

College of Arts & Sciences (CAS)

Discipline(s)

Data Science, Computer Science, Mathematics

ORCID Identifier(s)

0009-0003-0700-7889; 0009-0007-8399-7369; 0009-0008-7446-0719; 0009-0004-4247-9467;

Presentation Type

Poster Presentation

Symposium Date

Spring 4-24-2025

Abstract

Bitcoin was designed to be a decentralized peer-to-peer cash system, which creates scarcity in two ways: the total supply of Bitcoin available, and the block size of each block. The blocks, which form what is called the Bitcoin Blockchain, can contain at most 4 MB of data. Since the size of blocks is limited, and each transaction takes up some size, the number of transactions included in a block is necessarily limited. To combat this, users incentivize their transaction to be included in a block through a transaction fee, or the price the transactor pays a miner to include their transaction in a mined block. The miner, by validating the transaction and including it in a block can claim the fee as compensation. This price is typically quoted in satoshis per virtual byte or sat/vB. This will be the primary target unit for our fee estimation algorithm. What we propose is combining several data sources of current blockchain activity to predict the optimal fee. This approach is purely empirical, not relying on a theoretical model or explicitly estimating the time that the next block will be found. Additionally, we will also display the results of said algorithm to a monitor, utilizing MLFlow, so as to best visualize our results. This will enable us to better predict the costs of Bitcoin mining.

Biographical Information about Author(s)

Kristian Simakoski is a senior data science major with experience in financial analysis, statistical modeling, and business intelligence. Kristian has worked on projects analyzing NFL player performance, business cost evaluations, and data-driven decision-making.

Matthew Landrum is a sophomore at Valparaiso University studying Statistics, Mathematics, and Economics. He has also conducted research into Game Theory, acting as an author to “A Nash Solution to the Ukraine-Russia War”, which is set to be published soon.

Mason Ellerbroek is a senior with a Data Science and German double major and an interest in using data to better understand others and ourselves, particularly in language, trade, and inter-personal social dynamics.

Mason Tulacz is a junior Computer Science major who has interest in developing software solutions for various real-world problems. He has previously worked on projects making a citation generator, and an RPG video game.

Share

COinS