Exclusive Raid Invitations
Exclusive Raid Invitations

This research was made possible by Timmy Ma (Mathemagician10), a Mathematics student from the University of California. Timmy Ma used our public data set from our recent survey with over 5500 participants (link to data collection Google sheet) and his own (amazing) skills to provide us with a thorough regression analysis. 

The analysis, as you will see, leads to conclusions that are hard to dispute and could upset some of our readers — especially hardcore ones. 

A few days ago, we’ve published an article titled What the data tells us about this EX raid invitation cycle: February 8, 2018, in which we publicly shared our latest EX invites data set. This caught the attention of Timmy Ma, who decided to analyse the data set using a general logistic regression model and visualize his results using R.


Using a regression model where the y-axis is the prediction of receiving an EX pass using a general logistic regression model, the x-axis is the trainer level (see graphic below), Mathemagican10 was able to suggest the following:

  • the higher the trainer level, the lower the gym badge must be (at the targeted gym) in order to have a good chance of receiving an EX pass

In other words, the data suggests that for most Level 30+ Trainers it is easier to get an EX pass at a Bronze gyms than at a Gold gyms:

How to read the graph: for a level 40 player who is gold, they are predicted to obtain an EX pass at around 38%, but if that same trainer was bronze then their prediction would be around 68%. This is calculated from real world surveyed data and the regression model shows a very high degree of confidence that the trend will be preserved with more data.

Unfortunately, we can’t establish a trend for the lower level trainers, probably due to the fact that our voters include mostly trainers at higher levels.

Study details

  • 2908 randomly selected data points were used for this study
  • The p-value from the regression is < 0.00001
  • Analysis was tested on additional randomly selected data points, with no considerable effect on the trends.

Parting words

As Timmy suggests, more detailed data would need to be analyzed to confirm the theory, so we will make sure that this is reflected in our future surveys. There’s a ton of questions lingering around this:

  • Is the data set skewed and there are tons of casual bronze players who did not report?
  • Are reporting players doing “hit and run” raids at park gyms to cover any potential trigger?
  • Is the competition too high in the Gold / Silver badge levels, so the observed result is a product of this?
  • Is it possible that the level of randomness in the system causes results like these?
  • Will the results change when observing a single gym?

We don’t know at the moment, there’s simply too much confusion around the EX raid system. A huge thank you to Timmy for spending his free time on this analysis!

For more information on how to trigger an EX Raid read our guide here.