Forecasting Bitcoin Price With Crowdsourced Algorithms: Part II
TL;DR: This is the second article in my series on how I created an algorithm to predict the price of Bitcoin. In this article I am sharing all of my lessons learned and proposing a novel approach.
In my last article I created an algorithm that predicts the Bitcoin (BTC) price with a 29% investment return rate over 90 days. I shared that although I was initially impressed with my basic formula, I would not be satisfied until I was able to consistently repeat this performance over a longer period of time.
The simple formula uses a combination of price movement and the ratio of bitcoin related keywords indexed worldwide in Google Trends. My challenge was that the Google Trends data is a bit more unreliable on a longer scale since there is some kind of data smoothing or normalization over periods longer than 90 days. I was determined to get to the bottom of it.
Yet I have failed.
While working through this temporary setback I was able to connect with dozens of people that reached out wanting to help. Much ❤️ ️to my Medium followers! I spoke with a plethora of folks ranging from an astro physicist in the UK to a small hedge fund manager in Chicago to a blockchain and AI developer here in Taiwan. I learned that there are many others working on cryptocurrency price prediction and with various techniques. This is when it occurred to me that I didn’t need to go at this on my own. So an idea struck me — why not pool resources and learnings to tackle this challenge together?
Easier Said Than Done
The good news is that I’ve come across some wickedly clever approaches to crypto price prediction by many experts far smarter than me. There were algorithms that use sentiment analysis, technical analysis and other calculations that seemed very promising. The not so good news is that no one was satisfied with their individual results. Results were largely inconsistent across the board.
It dawned on me that until I can capture more consistent Google Trends data why not supplement with other data? I’ve collected and have started testing a small repository of datasets, backtesting strategies and quantitative approaches. What seems to be missing is a collaborative approach to share learnings and divvy-up expertise. This is why I’m building the AlgoHive project, a crowdsourced hive for predictive algorithms focusing on cryptocurrency.
What seems to be missing is a collaborative approach to share learnings and divvy-up expertise.
Welcome to the Project AlgoHive Community
There are many great teams that are already working on prediction markets, AI micro-service marketplaces and cryptocurrency portfolio recommendations. The demand for better predictive models are certainly there but suppose we could build a platform that incentivizes the algorithm builders, crypto speculators and potential investors while at the same time publicly sharing lessons learned? This is the vision for Project AlgoHive and we’re now building-up the team.
At my previous Fintech startup, an AI-based virtual advisor for helping people avoid debt from medical emergencies, I’ve learned a lot about how challenging it can be to build a smart recommendation algorithm. Building a predictive algorithm certainly ups the game. Having great tech is paramount but even more important is getting the right people on the bus. This bus has now arrived and I hope you’ll join me for an exciting trip.
In subsequent posts I’ll be sharing the details on what I envision as an MVP (Minimal Viable Product) for the AlgoHive platform. I’ll also share how this platform’s multi-pronged incentive system can be distributed to all project stakeholders and governed via a smart contract.
Here’s what you can expect should you decide to get on the 🚌 and join our budding community:
- Meet and share ideas with some of the best minds in the areas of machine learning/artificial intelligence and blockchain development.
- Learn more about cryptocurrency investment strategies and what signals that potentially impact markets.
- Access to a growing repository of datasets, APIs, algorithms and sentiment analysis sources that can help towards crypto market price prediction.
- Participate in the guidance and testing of the AlgoHive platform MVP.
- Follow along the performance of these algorithms and team strategies.
- Become an early backer in the first crowdsourced platform for cryptocurrency investment algorithms.
To follow along please follow me here on Medium then join our growing AlgoHive Discord community! If you’re interested in joining the team or becoming an advisor let’s chat.
All aboard…to the moon! 🌕
This article is intended for informational purposes only. The views expressed in this article are not, and should not be construed as, investment advice or recommendations.