Financial Asset Statistical Analysis
and Price Forecasting


Project Overview

Financial time series forecasting is challenging for a multitude of reasons, particularly due to the highly stochastic nature of the dynamical system. However, traditional assets have been well studied and have petabytes of data in order to develop forecasting models. Unlike traditional assets, cryptocurrencies are still relatively new and have limited amounts of data in comparison. Quantum Analytica partnered with CoinGenius to develop a novel approach to cryptocurrency forecasting that blended traditional financial forecasting methodologies, such as Markovian models, with non-linear dynamical methods and deep learning to develop a novel model pipeline that could robustly forecast daily and weekly trends in cryptocurrency market.

Our Deliverables

Exploratory Data Analysis

Data Visualization

Advanced Market Analytics

Cryptocurrency Price Forecasting at a minimum of 65% accuracy

Project Goals

Cryptocurrency markets are highly volatile and those who trade in this asset class have a wide bifurcation of skill sets; some are hobbyist traders, while others are expert quants from Wall St. CoinGenius recognized that analytics and modelling in the crypto market lacked robustness, accuracy, and was not accessible to traders of a wide variety of skillsets. Quantum Analytica partnered with CoinGenius to develop a highly sophisticated forecasting algorithm and real time analytics in order to give traders, no matter their background, the strongest insights in order to make data-driven trading decisions.

Our Approach

We developed custom ETL and data normalization pipeline in order to ensure data validaty, and low latency, across a multidude of traditional and crypto exchanges. Utilizing Bayesian inference, time series and spectral analysis, and Markovian models we developed a suite of analytics that related traditional market assets to crypotcurrencies. Because cryptocurrency is relatively new, we had to develop a novel deep learning pipeline that blended methods from non-linear dyanmics, chaos theory, spectral analysis, and generative models in order to provide additional contextual features to the final model. This approach allowed us condition our models on the features that mattered, allowing us to meet the accuracy requirement.


Client Feedback


“Russ is one of the most incredible data scientists I have ever had the pleasure of working with. He is able to take complex data sets and extract valuable insights using a variety of advanced AI and machine learning techniques. His ability to innovate and push the boundaries of what is possible is a unique skill that I seldom come across in this field and has led to several breakthroughs in our organization.

Jeremy Born

CEO, CoinGenius