Greetings, friendly travelers! The Pokémon GO Lunar New Year Event is underway, lasting from January 24th until February 3rd. During this time, Silph Researchers have been gathering with friends and collecting data for all types of Best Friend interactions. Our Researchers are on the road to determine the Lucky Friends rate for this event! Because the Lucky Friend feature wasn’t available during last year’s Lunar New Year Event, this event was uncharted territory for our Researchers. Here’s what we have discovered so far!
2/2/2020 22:00 UTC Update
Since the event began, researchers have completed 1,894 Best Friend interactions and acquired a total of 66 Lucky Friends, giving a 99% confidence interval around the Lucky Friends rate of 2.49% to 4.66%. Compared to the standard (non-event) Lucky Friends rate, around 1%¹, these initial results suggest that the rate of Lucky Friends is potentially three or four times the unboosted rate.
When gathering data on rare events, reporting bias is always an issue. Reporting bias comes from the tendency to report memorable events more reliably than uninteresting ones. For example, it is easier to forget to report an impromptu trade if it is unremarkable. One way that we have found to detect bias in our data sets is by comparing our more experienced researchers to the researchers still learning the ropes. Senior members may enter data manually, while more junior members use a form. This separation in the data set is a simple way to check for differences between the two groups.
In this data set, we find that experienced, senior researchers report lower rates of Lucky Friends compared to other researchers. The substantial (albeit statistically non-significant) differences in reporting types suggest that the actual rate may be closer to the 3% rate reported by senior researchers.
|Interactions||Lucky||Rate||99% confidence interval|
|Form Data (newer researchers)||816||34||4.2%||2.6 – 6.2 %|
|Manual Data (experienced researchers)||1,078||32||3.0%||1.8 – 4.5 %|
There are other alternative hypotheses as to why these two groups may differ, but reporting bias seems to be the simplest explanation.
Because of the change we detected in Lucky Trade rates after the first day of the event, we were anxious to see if similar trends could be detected for the Lucky Friends rate. While the limitations of available Best Friends means that data per day is limited, our analysis does not suggest large fluctuations over time. (Dates are reported in local time at the time of the interaction.)
|January 24||251||9||3.6%||1.3 – 7.3 %|
|January 25||306||15||4.9%||2.3 – 8.6 %|
|January 26||296||10||3.4%||1.3 – 6.7 %|
|January 27||234||11||4.7%||1.9 – 9.0 %|
|January 28||235||6||2.6%||0.7 – 5.9 %|
|January 29||191||6||3.1%||0.8 – 7.3 %|
|January 30||175||6||3.4%||0.9 – 7.9 %|
In some languages, the Lunar New Year Event announcement suggested that trade interactions may have a higher rate of triggering Lucky Friends than other interaction types. While Niantic Support responded on Twitter to confirm that the rate should be equal across interaction types, the question remained. Trade data has been less abundant thus far, but the initial analysis suggests that trade interactions likely do not have a higher rate of triggering Lucky Friends than other interaction types.
|Interaction Type||Interactions||Lucky||Rate||99% CI|
|EX Raid Invite||8||0||–||–|
|Gift Opened||1,404||55||3.9%||2.7 – 5.4 %|
|Trade||198||6||3.0%||0.8 – 7.0 %|
|Trainer Battle||232||5||2.2%||0.5 – 5.4 %|
Researchers will continue to track their interactions with Best Friends throughout the event. We will keep you updated as we continue collecting and analyzing the data! Until then, stay safe on the Road, travelers!
¹The base rate of Lucky Friends is currently under extensive study by Silph Researchers. Be on the lookout for a comprehensive article as soon as the study is completed! 1% is a rough estimate and should not be considered a definitive statement about the base rate of Lucky Friends.