Counter-Strike : Global Offensive (CS:GO) is a popular first-person shooter video game with a unique feature: its out-of-game economy. CS:GO players purchase virtual cosmetic items called “skins” that decorate their weapons in-game yet provide no functional utility. Despite this, skins frequently sell for hundreds of US dollars. This paper investigates the nature of the CS:GO skins market and asks whether skins are more similar to consumer goods or more similar to speculative virtual assets like cryptocurrencies. I summarize the literature on virtual item pricing as well as literature on price prediction of cryptocurrencies. In order to assess the skins market quantitatively, I implement tree-based machine learning models to estimate the price of CS:GO skins based on their observable attributes and based on their historical price data. The attribute-based models reveal that observable attributes have little predictive power, while the historical price data models perform extremely well. I argue that because there is no clear relationship between skin attributes and their prices and because skin prices are well-described by methods used in cryptocurrency price prediction, CS:GO skins might be more like virtual commodities than traditional consumer goods.
Metrics
81 File views/ downloads
238 Record Views
Details
Title
An investigation on the pricing of virtual items & digital commodities
Creators
Emma Taylor
Contributors
Alejandro Prera (Advisor)
H Alan Love (Committee Member)
Jia Yan (Committee Member)
Awarding Institution
Washington State University
Academic Unit
Economic Sciences, School of
Theses and Dissertations
Master of Science (MS), Washington State University