Knowledge Center

Video Introductions, Examples and Structured Learning

Welcome to the prediction network.

To motivate this project, here's a first look and some use cases to get you thinking about the potential for an open, democratic prediction network. If nothing else it automates benchmarking, and offers a new way to defining time-series anomaly, but it might also bring AI within the reach of individuals and small business some day.

To that end we produce open-source code that and is fast and simple, and intended to foster collective AI at web scale. We run daily and monthly competitions, and we're also sponsoring the M6 Financial Forecasting Competition that will run throughout 2022 with $300,000 in prizes. We have a bit of fun with this, so join our slack channel.

Join Our Slack Channel
Peter Cotton

Virtual Office Hours

Jump-start your participation by joining us for one of our collaborator discussions, hosted by site developer Peter Cotton. 

Tuesday 8:00-9:00pm EST
Friday 12:00-1:00pm EST 



It moved to Frequently asked questions. Raise an issue on Github. Email us if you are stuck or jump on our Slack.

Python Ultra-Quick Start

There's a new way to enter contests that is dead easy and "set and forget". Just fork this repository and enable GitHub actions. There's a notebook in the repo you can use to generate yourself an identity. There are some limitations to this approach, but it will get you on the leaderboards very quickly. See the article How to Enter a Cryptocurrency Copula Contest.

Python Video Tutorials

To create a "crawler" that continuously predicts streams, it is best to work through these tutorials covering the basics.   

  1. Your first submission
  2. Your first crawler
  3. Retrieving historical data 
  4. Creating a data stream
  5. Modifying how your crawler makes predictions
  6. Modifying where your crawler makes predictions

Python Notebook Examples

The repository contains notebook examples you can easily run (for example just open them in Google Colab).

and more. 

More Python Patterns  

You can love or leave 

  • FitCrawler  (Comal Cheetah) re-estimates models during downtime. It illustrates some abstracts that might drastically shorten your path to creating good crawlers, notably:
  • GitHub actions enable free ongoing scheduled computation. This is a great pattern for participation when a running process (crawler) is overkill, or when on-the-fly model fitting is unwieldy. With GitHub actions you can:
    • Maintain a repository with up to date model parameters like these expnorm files. 
    • Top up your balance, as with the key maker.
    • Submit predictions for z-streams, as with this example.
    • Generate reports and pretty pictures.
  • Bouncing using bash scripts, or otherwise, to keep your crawler going. 

More on the way. 


Welcome R statisticians! 

More on the way soon. 

Julia Examples  

Welcome, Julia developers!  Video tutorials are on the way. For now:

More at Rusty's repo including an advanced electricity prediction project

Player Strategy

Be sure you understand the mechanics of how predictions are quarantined and rewarded.

Business Context

For those who may be interested, and willing to invest the time, here is a sequence of talks providing what is hopefully a coherent case for an open prediction network. It isn't an elevator pitch.

Along the way, we explain why vendor Automated Machine Learning and previous attempts at crowdsourcing might violate a fundamental axiom, at least if quantitative business optimization is to be fully democratized. Pro tip: Some people like to start with #3. 

  1. A first look at microprediction and how it gives rise to the "ten minute data science project". 
  2. What must a microprediction oracle do?  The answer isn't Kaggle, or DataRobot. 
  3. Business uses of microprediction include pretty much everything (unless "prediction" is in the name, ironically). 
  4. Repeated value function prediction provides the link between microprediction and Control Theory and Reinforcement Learning, which is why microprediction is truly general. 
  5. Why algorithms make excellent managers, and by implication why humans are terrible. Copious theory means algorithms are well placed to orchestrate production of prediction. Humans need not occupy a blocking role.

Learn Data Science

If one person decides to learn to code so they can use the prediction network, we're done. Seriously it isn't that hard. Here are some entirely free ways to get going.

  1. Visual Coding and games to get started with coding concepts (my daughter and I love some of these).
  2. Learn to Code, in Python if you've never coded before.
  3. Learn Python more quickly if you've coded in another language. 
  4. Learn data science using Python. 


Please contact us if you can help improve these resources. As noted above we'd love to answer your questions via Gitter, or issues on Github, the discussion on GitHub, or even questions at Quora. You can also email us directly. 

Follow the Movement