Trading is a staple feature in Pokémon games and became an integral part of the Pokémon GO universe in July 2018. Pokémon GO introduced a twist to trading compared to the main series games: Pokémon reroll their stats when traded. Travelers often take advantage of this by trading away their Pokemon with subpar IVs in hopes of rerolling better ones. One myth that emerged shortly after the release of trading was that Pokémon with higher stats are more likely to keep high stats after trading. To many travelers, this seems intuitive. A Pokémon with higher stats should be more valuable in a trade than one with lower stats, right?
To put this myth to the test, researchers in the Silph Research Group got together with friends and conducted over 3,000 trades in a controlled study. After careful analysis of Pokémon stats before and after trading, we’re ready to call this myth:
Each Pokémon has three stats (Attack, Defense, and Stamina, referred to as Individual Values [IVs]) that are added to the base stats of the species. IVs range from 0-15 for each single stat and are individually assigned to each Pokémon. When a Pokémon is traded, its IVs are randomly changed.
In a Developer Insights post released shortly after the trading feature launched, Niantic provided the following details regarding trade mechanics:
Another notable aspect of trading in Pokémon GO is that the stats of the Pokémon will change randomly within a range when traded…. We first look at your friend’s Trainer level to confirm the Pokémon is not powered up past what would ordinarily be obtainable for someone at that level. Next, we assign minimum stat values based on your Friendship level, and that base increases the higher that level is. – pokemongolive.com
Based on these details, the Research Group hypothesized that the original IVs of a Pokémon do not affect the IVs after it is traded.
For each trade, researchers recorded the trainer levels of the trade participants, the friendship level with the trading partner, the appraisals before and after the trade. Factors such as friendship level and lucky status of a trade impact what the minimum value of each IV can be after trading.¹ For example, if a trainer is trading with a Good friend, the minimum IVs possible are 1/1/1. This is referred to as the IV floor. If the trade goes lucky, the minimum IVs possible are 12/12/12, regardless of friendship level. Collecting information on the friendship level of the trading partners, along with the lucky status of each trade, allowed us to exclude unobtainable IV combinations from our expected counts for appraisals.
To test our hypothesis, we first looked at the post-trade IV distribution. The total IVs for each traded Pokémon was categorized into a percentile range using the same boundaries as the “star” system of the Team Leader appraisal.² Using a Pearson’s Chi-squared (χ2) goodness-of-fit test, we compared the number of traded Pokémon observed in each category to the number of Pokémon we’d expect to fall into each category if the IVs were uniformly distributed. The table below shows the observed IVs post-trade, as well as the number of Pokémon we’d expect to fall into each category if each IV combination was equally likely (given the IV floor). We’ll highlight Ultra and Best friend trades, which have the most data points.
||Post-trade Count||Expected Count|
||Post-trade Count||Expected Count|
For all friendship IV floors, we were unable to reject the null hypothesis at the chosen significance threshold of 0.05. This means that a Pokémon’s IVs following a trade closely align with a uniform distribution, and do not trend toward a particular star appraisal.
The Relationship of Traded IVs
The next step was to examine the relationship between a Pokémon’s IVs before and after a trade. The plot below shows how the sum of the three IV values changes for a single Pokémon from before the trade to after the trade.
Relationship between the total IVs of a Pokémon prior to trading and the total IVs after trading. Each point is a single traded Pokémon. A small amount of noise was added to each point to help with visualization. The points are colored according to their IV floors, based on friendship level and lucky status. The black dashed line is the best fit simple linear regression of pre- and post-trade total IVs. The model explains only 0.1% of the data and is not significant.
To formally test the relationship between pre-trade and post-trade IVs, we used a multiple linear regression. The IV floor (Good, Great, Ultra, Best, and Lucky) of the trade was added to the model along with total IVs pre-trade (F(2,3514) = 464.3; R²adjusted = 0.209). Of the individual variables, IV floor was a significant predictor when determining post-trade IVs (t = 30.384, p < 0.001), whereas pre-trade IVs had little to no impact on post-trade IVs (t = 1.317, p = 0.188).
In addition, correlations between the Attack, Defense and Stamina stats before and after trading were examined separately. No significant correlation existed between the pre- and post-trade single stats.³ For avid GO Battle League participants, this means that a Pokémon with a low Attack stat is no more likely to have a low Attack post-trade than a Pokémon with random IVs. Other than the IV floors guaranteed by friendship level, it seems the post-trade IVs are best explained by RNG!
Many thanks to our Senior Researchers who collected data on thousands of trades to show that there is no significant correlation between the pre-trade and post-trade IVs. Travelers can take satisfaction in knowing that their rerolled IVs will be completely random. Trading away unwanted Pokémon remains the best strategy to turn those trubbish IVs into treasure! Until our next trade, travelers, see you on the road!
Article author: Scientist CaroKann
Analysis: Scientists CaroKann and Titleist
Editing: Scientists Cham1nade, skyeofthetyger, Titleist, and Lead Researcher archer
Graphics: Scientist WoodWoseWulf and Titleist
Project Leaders: Scientist Mihryazd
¹ The minimum IV for each trade category is shown below.
|Trade Category||Minimum IV|
² A majority of the data was collected using the old IV appraisal system, which relied on team leader phrases to estimate IVs. Under the old appraisal system, the first phrase was used to indicate the “star” (0* – 4*), and the second phrase was used to indicate the the highest single IV value within a predetermined range. Using a combination of the team leader phrases, Pokémon species, CP, and HP values, an in-house algorithm was used to find the correct IV combination. In the event multiple IV combinations fit the criteria, the mean value of all possible IVs was assigned.
³ Using Pearson’s product-moment correlation, we compared each single IV before and after a trade. I.e. the pre-trade Attack Stat was compared to the post-trade Attack Stat. No significant correlation was found between the pre-trade and post-trade IV stat for Attack (r(3515) = 0.025, p = 0.14), Defense (r(3515) = 0.018, p = 0.28), and Stamina (r(3515) = 0.014, p = 0.41).