Research

Cryptocurrency Forecasting Model

San Jose State UniversityDecember 2022

Abstract

The cryptocurrency exchange domain is a relatively volatile space. The most widely traded cryptocurrency coin Bitcoin has experienced a high of $44,533.00 and a low of $36,259.01 in the week of 1/31/22 - 2/7/22. The volatility of the cryptocurrency market stems from three accepted analyses. A technical analysis solely relies on metrics ranging from historical trends to net unrealized profit/loss to derive the effects of price movements. A fundamental analysis relies on factors that affect price movements, such as government policies. A sentimental analysis relies on the sentiment of a coin at a particular time, which can be identified using social media trends. Given the abundance of variables that affect price movements, forecasting even near-future prices prove difficult for many traders. Each of the three analyses stated (technical, fundamental, and sentimental) have sub-analyses that would take an abundance of time even for the experienced trader. As the digital asset market increased exponentially over the past 2 years, many traders are not accustomed to these analyses, much less able to derive conclusions from them. The cryptocurrency forecasting model aimed to traverse, analyze, and interpret data from the three types of analyses with a greater focus on technical and sentimental analysis. Using the data interpreted, the model has the ability to forecast price movements to the time scale of the customer's preference. This project reduced the time spent significantly analyzing technical data, assisted traders to make confident trading decisions, and detailed the price movement patterns that are difficult to infer with purely human capabilities.

Keywords

Machine LearningCryptocurrencyNeural NetworksNLP

Preparing document

Cryptocurrency Forecasting Model