DEV Community

Gabriel Henriques Gomes
Gabriel Henriques Gomes

Posted on • Edited on

1

Forecasting cryptocurrencies prices using social community historical time-series | Introduction

The following article describes a Data Science project Introduction.
It is followed by every step, from data gathering to final insights over cryptocurrencies price action behavior, each with a dedicated article.


The Problem

Asset price valuation 
Predicting an asset's price action behavior has been a tempting goal in investment science for a while, and never an easy one. 

For crypto-assets, practical intangibility extrapolates valuation methods subjectiveness even further, encouraging the need for new points of view.

Metcalfe's law solution approach

Metcalfe's law, one of these points of view, is not new at all: conceived in 1983 in a presentation to the 3Com sales force, it stated that the value of communication technologies (specifically the Ethernet) would be proportional to the total number of possible connections, or value=a*n^2 (a = value per connection | n = users).

In 1995, impressed by the range of applications for Metcalfe's 'principia', the venture capitalist George Gilder named it Metcalfe's Law. If you'd like to hear the words from the man himself: click here
 
The conceptual proof came only in July 2013, when Dutch researchers managed to analyze European Internet usage patterns over a long enough time and found n^2 proportionality for small values of n, and n log(n) proportionality for large values of n, solidifying what is today known as the 'Network Effect'.

A few months later, Metcalfe himself provided further proof, as he used Facebook's data over the past 10 years to show a good fit for Metcalfe's law (the model is n^2).

What is new is that only 3 years ago, Peterson linked time-value-of-money concepts to Metcalfe value, using Bitcoin as numerical examples of the proof and showing that over 70% of the variance in Bitcoin value was explained by applying Metcalfe's law to increases in Bitcoin network size.

Impact 

Peterson's work has not only contributed to the statement of Metcalfe's law, but by using Bitcoin's price action to do it, he also unleashed the concept of mathematical proof for cryptocurrencies pricing models

Metcalfe's approach limitations

An old critique (Odlyzko, 2006) states that Metcalfe is wrong when assuming that the value of each node (n, user) is of equal benefit, and that the applicability of his law is constrained to 'long enough' time-series, as patterns may average-out utility discrepancy. 

So with this project, we will try to find a composition of network metrics (a composed user) that serves as an indicator of the future performance of a given crypto asset. 


Project objectives

Inspired by the 'Network Effect' from Metcalfe's Law, this project aims to retrieve insights from time-series analysis over different crypto tokens 'Users'.

We will collect 3 years worth of daily data, for 50 different tokens, for the following metrics:

Social community data
'twitter_followers',
 'reddit_subscribers',
 'reddit_average_posts_48h',
 'reddit_avg_comments_48h'

Developers community data | cryptocurrency´s github repository
'forks',
 'stars',
 'github_subs',
 'total_issues',
 'closed_issues',
 'pull_rqst_merged',
 'pull_request_contributors'

Market_cap, total_volume, current_price
In USD, EUR and BRL


By the end, we will be in the position to answer the question:

'Does any of the considered variables (or any conjuncture of them) help to estimate the 'fair-value' and 'future price' of a cryptocurrency?'


Analytics Engineering | Technologies 

The workflow will be hosted by Google Cloud Platform services. 

Cloud processing and Storage
The database will be stored in Google Storage, updated through a python script scheduled with cron jobs, which runs in a scheduled preemptible virtual machine from Cloud Compute Engine.

Database Distribution
A Google Cloud Function will also be triggered everyday to transfer updated data from Google Storage to Google BigQuery, making it accessible through DataStudio and Tablaeu.

Top comments (0)

👋 Kindness is contagious

Please leave a ❤️ or a friendly comment on this post if you found it helpful!

Okay