This website shows real-time predictions of the success of Kickstarter campaigns. These predictions are regularly updated to take into account the latest evolution of the campaigns. They should be taken as purely speculative: there is no guarantee of their accuracy, and they are subject to change radically at any moment.


The predictions are obtained by mixing several individual predictions, that are each made using different sources of information:

  • the amount of money pledged over time
  • tweets containing a link to a campaign
  • backers and their relationships with each other

If you want more details about how the predictions are made and what exact features are used, please have a look at the related paper: Launch Hard or Go Home! Predicting the Success of Kickstarter Campaigns.

On a more technical side, this website uses Twitter Bootstrap on the frond end, and Zend Framework on the back end. The predictors are implemented in Python and some use Scikit Learn.


My name is Vincent Etter. I am a PhD student from EPFL in Lausanne, Switzerland. I am not related in any way to Kickstarter, nor do I claim ownership of any of the data used or displayed on this website. I decline any responsibility with respect to the success or failures of campaigns: my predictions are purely informational, and should not be used as basis for taking any decision.


I built this website to showcase some of the work I did during my PhD, and to share the data I used to do it. I really love the idea behind Kickstarter, and thought that it would be interesting to see wether we could predict the success of the campaigns. So I tried!