Pokémon Go Spawn Location Analysis

Tom Mathews | Darkphibre

July 25, 2016

Abstract

A dataset spanning 110 hours, generated from a variety of Pokémon monitoring stations, was provided for analysis. Pokémon appear to spawn at fixed locations in the world, once an hour, and run away after 15 minutes. Pokémon ‘density’ is directly correlated to the number of spawn locations within view: In locations with spawning Pokémon, 40% will have on average one Pokémon visible at a time. Human population is strongly correlated with Pokémon spawn rates, culminating in regions with the lowest human population having Pokémon spawn rates of 0%.

Process

Note: This research is meant to provide insights into the lives and habitats of Pokémon, and is not intended to convey inappropriate advantages to Pokémon poachers. As such, specific details of the location of the monitoring stations will be kept private, and the researchers involved in this project are refraining from jumping in the car and heading to the nearest nest [if one were to exist]. A total of 160 stations were placed within the world, each separated by .002 latitude / longitude. We were assured that the sampling was done in an intermittent fashion to avoid disturbing the population. The coverage was selected to identify metropolitan, urban, rural, and lake habitats. Of those 160 stations, only 96 stations actually observed one or more Pokémon. Intermittent sampling provided observations on 41,657 distinct Pokémon over the course of 110 hours.

Findings

We will discuss our findings, starting with individual spawn locations and moving into more general summaries.

Distribution of Pokémon Locations / Distance from Observer

Below is a distribution plot of the Pokémon observed, combining all observation stations. As you can see, wild Pokémon are not observable outside a distance of 100m from the observation point (denoted by the red circle). Instead of Observation Posts, you can think of yourself standing in a field. How far can you see pokemon? How many will spawn around you during the course of an hour? That is what we’re talking about.

Of particular note is the sparsity of locations, the small number of black dots. Each of the 422 points represents 99 individual Pokémon spawns occurring in exactly the same location. If Pokémon spawning was random, this should be an evenly-distributed cloud of points. As this is not the case, it’s clear that Pokémon spawn in fixed locations.

Finally, the individual spawn locations themselves are relatively evenly distributed, with a slight increase in density at the diagonal directions. This merits further investigation with additional observation locations.

Spawn Location Distance from Observation Station

Spawn Location Distance from Observation Station

Variety of Pokémon Spawns within Station View

Given that Pokémon spawn in a fixed location, the next question becomes: How many spawn locations are observed by a typical monitoring station? Grouping each monitoring station based on the number of spawn locations contained within the monitoring station reveals that most stations have access to five or fewer Pokémon spawn locations, with very few having access to more than 10:

Distribution of Visible Spawn Counts

Distribution of Visible Spawn Counts

Time To Spawn

It’s become common knowledge that the act of a Pokémon spawning at a specific spawn location occurs on an hourly basis, and this dataset seems to confirm this observation1 This is not charted, as it’s a rather boring chart. Every single one of the 41,657 Pokémon was observed for no more than 15 minutes. The ones that were observed less than 15 minutes were during the start or end of the observation period.,2 Interesting! I’ve been shared data that seems to indicate there are 30-minute spawns. I’ve double-checked my data, and the only time this happens is when the servers seem to have a hiccup. That said, I may have more information in the next week, so stay tuned!. Hourly spawning of Pokémon is only part of the story, as an observer in any given location may have access to additional spawn locations.

Turning our gaze to these observation stations, one could think of it as standing in a singular place: What is the average time you would have to wait until another Pokémon arrived (of any type), given the time of the last Pokémon spawn? This is strongly correlated to the number of spawn locations an observation station has access to.

In referring to the cumulative frequency chart below, you can determine that in this dataset it would be 50% probable to see a new Pokémon within 18 minutes while standing stationary.

Observation Stations: Mean Minutes Until Spawn Each bar in this chart represents the number of stations that have an average time-until-next-spawn of N minutes.

Observation Stations: Mean Minutes Until Spawn
Each bar in this chart represents the number of stations that have an average time-until-next-spawn of N minutes.

Given that Pokémon spawn every 60 minutes and persist for 15 minutes, observation stations with 4 or more spawn points become increasingly likely to have concurrent Pokémon. The sampled probability of coming across multiple Pokémon was found to be:

Pokemon Seen Concurrently Probability
1 74.659%
2 18.743%
3 04.354%
4 01.371%
5 00.541%
6 00.182%
7 00.065%
8 00.055%
9 00.020%
10 00.003%
11 00.007%

If one were waiting for a specific Pokémon to respawn, that is another story. Each spawn location may spawn a variety of Pokémon, and so anything but the most common Pokémon will take several hours to spawn at the same location. Please refer to Pokémon Spawn Frequencies per Station for a list of Pokémon type / Pokémon respawn probabilities.

Sparsity of Rural Areas

It is interesting to note that the sparsity of spawning as a station becomes more rural is extremely significant. In the diagram below, the gaps in Pokémon spawning correlates to:

The northern portion of the map contains a bustling city, and the triangular quadrant mid- to mid-west are trails cutting through the large park system (leading to the lake). You can see that low-population areas have had no Pokémon spawning during the observation period of 110 hours:

Pokemon Spawn Locations, and Observation Posts

Pokemon Spawn Locations, and Observation Posts

Future Work

Future research includes plotting the distribution of more rare Pokémon over time, identify what else is spawning when rare Pokémon are not spawning, and the probabilities of various effects and how they pertain to spawn probabilities.

The effect of Lures on Pokémon spawn rates, both for those lured and wild Pokémon spawning nearby, was explicitly removed from this analysis, and merits additional research.

Finally, the concept of ‘nests’ is an area that merits research. While not explicitly searched for, there certainly seems to be several monitoring stations with a propensity for multiple concurrent spawns of various creatures. Refer to bars with more than 10 spawns in the Time To Spawn chart.

About the Author

Tom Mathews has been an avid videogame player since the Atari 2600. He’s spent the last 20+ years developing data-driven systems, including 5 years at the Advanced Technology Group in Xbox and the last 5 years at 343 Industries helping to architect cutting-edge big data acquisition systems [see his GDC 2016 talk for details]. Over the last year, he’s been focusing more on the ‘big data’ analytical side of equation, helping to identify interesting patterns in large datasets to improve gameplay and component interactions. He’s been fascinated by the Pokémon Go cultural phenomenon, having hiked over 60KM the first week hatching eggs and hunting Pokémon. He rarely speaks of himself in the third person, and is wondering if this is overding things.3 Nah.

This paper reflects his own opinions, and do not reflect those of his employer.

No Pokémon were intentionally harmed during this research.

Appendix

Pokémon Spawn Frequencies per Station During 30 Hours Observation

Requires two observations in the same station. This window was chosen as one large enough to contain at least 24 hours of consistent data.

Observations of Pokémon Repeat Spawns

Type Repeat Observations Median Hours Between Spawns Average Hours StDev Hours
Any Pokemon 18,495 0.4 1.1 2.4

Observations of Pokémon Repeat Spawns by Type

Type Repeat Observations Median Hours Between Spawns Average Hours StDev Hours
bug 4,032 0.4 0.9 1.4
dragon 1 46.0 46.0 NaN
electric 13 8.5 9.5 7.2
fairy 348 1.9 3.4 4.3
fighting 39 4.2 7.3 8.1
fire 16 11.1 13.0 10.1
ghost 151 4.8 6.7 7.3
grass 392 2.2 3.9 4.8
ground 41 3.6 7.1 8.0
ice 3 14.0 14.0 11.1
normal 7,204 0.3 0.5 0.7
poison 2,954 0.5 1.0 2.1
psychic 581 1.5 3.1 4.4
rock 11 13.0 14.5 13.2
water 2,709 0.6 1.1 2.0

Observations of Pokemon Repeat Spawns by Species

ID Pokemon Repeat Observations Median Hours Between Spawns Average Hours StDev Hours
7 SQUIRTLE 40 6.0 9.5 10.4
10 CATERPIE 600 1.7 3.5 4.8
11 METAPOD 24 8.2 9.6 9.3
12 BUTTERFREE 1 8.2 8.2 NaN
13 WEEDLE 2,162 0.9 1.5 2.2
14 KAKUNA 92 8.4 9.7 7.6
15 BEEDRILL 6 10.3 11.3 6.9
16 PIDGEY 2,894 0.6 1.2 1.7
17 PIDGEOTTO 129 4.9 7.4 7.5
18 PIDGEOT 5 12.5 16.7 8.7
19 RATTATA 1,255 0.9 1.8 3.3
20 RATICATE 23 8.0 10.8 9.0
21 SPEAROW 1,425 1.1 2.2 3.2
22 FEAROW 15 10.9 12.0 9.7
23 EKANS 1 10.1 10.1 NaN
25 PIKACHU 1 2.8 2.8 NaN
27 SANDSHREW 19 2.0 2.4 1.6
29 NIDORAN-F 388 2.0 3.9 5.2
30 NIDORINA 10 10.8 11.7 9.2
32 NIDORAN-M 407 2.1 4.2 5.7
33 NIDORINO 4 19.0 18.8 5.0
35 CLEFAIRY 342 2.0 3.4 4.4
39 JIGGLYPUFF 74 5.2 7.6 7.0
41 ZUBAT 1,860 0.6 1.3 2.4
42 GOLBAT 40 8.0 10.5 8.9
43 ODDISH 157 4.0 6.0 6.4
46 PARAS 260 3.3 6.3 7.4
48 VENONAT 410 3.1 4.7 5.1
49 VENOMOTH 1 15.6 15.6 NaN
52 MEOWTH 50 5.8 9.4 9.1
54 PSYDUCK 48 3.1 4.9 4.4
55 GOLDUCK 1 32.9 32.9 NaN
56 MANKEY 7 9.3 10.3 8.1
58 GROWLITHE 6 23.5 22.5 6.0
60 POLIWAG 47 3.7 4.5 3.7
63 ABRA 11 6.9 8.6 8.8
64 KADABRA 1 5.6 5.6 NaN
66 MACHOP 17 3.9 6.5 7.8
69 BELLSPROUT 156 4.4 6.4 6.6
70 WEEPINBELL 1 6.9 6.9 NaN
72 TENTACOOL 5 4.2 9.4 13.5
74 GEODUDE 8 14.9 15.6 13.1
79 SLOWPOKE 26 8.0 8.5 5.7
81 MAGNEMITE 2 17.8 17.8 2.5
86 SEEL 38 7.5 10.6 10.1
88 GRIMER 1 26.3 26.3 NaN
90 SHELLDER 1,682 0.6 1.3 2.8
92 GASTLY 139 5.1 7.0 7.5
93 HAUNTER 3 12.0 10.0 8.5
96 DROWZEE 534 1.6 3.2 4.5
97 HYPNO 2 6.1 6.1 2.8
98 KRABBY 199 4.0 6.9 8.0
104 CUBONE 1 8.0 8.0 NaN
109 KOFFING 9 13.5 15.3 8.6
111 RHYHORN 12 16.1 16.8 12.0
116 HORSEA 70 5.4 8.1 8.7
117 SEADRA 1 0.3 0.3 NaN
118 GOLDEEN 50 4.0 5.7 6.2
119 SEAKING 2 11.4 11.4 2.0
120 STARYU 37 5.1 8.2 7.3
123 SCYTHER 117 0.5 1.4 2.0
124 JYNX 3 14.0 14.0 11.1
125 ELECTABUZZ 7 5.0 6.7 3.6
129 MAGIKARP 127 2.0 3.3 4.0
133 EEVEE 872 1.4 2.9 4.3
147 DRATINI 1 46.0 46.0 NaN

Total Pokémon Spawn Frequencies

species_id Pokemon Observations
1 BULBASAUR 15
4 CHARMANDER 9
5 CHARMELEON 2
7 SQUIRTLE 167
8 WARTORTLE 16
9 BLASTOISE 1
10 CATERPIE 1,452
11 METAPOD 95
12 BUTTERFREE 12
13 WEEDLE 4,990
14 KAKUNA 335
15 BEEDRILL 53
16 PIDGEY 6,495
17 PIDGEOTTO 409
18 PIDGEOT 39
19 RATTATA 2,808
20 RATICATE 105
21 SPEAROW 3,287
22 FEAROW 102
23 EKANS 29
25 PIKACHU 18
27 SANDSHREW 55
29 NIDORAN-F 954
30 NIDORINA 59
31 NIDOQUEEN 3
32 NIDORAN-M 1,013
33 NIDORINO 41
34 NIDOKING 2
35 CLEFAIRY 841
36 CLEFABLE 14
37 VULPIX 20
39 JIGGLYPUFF 245
40 WIGGLYTUFF 2
41 ZUBAT 4,114
42 GOLBAT 153
43 ODDISH 452
44 GLOOM 23
46 PARAS 722
47 PARASECT 23
48 VENONAT 1,042
49 VENOMOTH 37
50 DIGLETT 18
51 DUGTRIO 2
52 MEOWTH 182
53 PERSIAN 4
54 PSYDUCK 146
55 GOLDUCK 7
56 MANKEY 76
58 GROWLITHE 39
60 POLIWAG 140
61 POLIWHIRL 10
63 ABRA 72
64 KADABRA 6
66 MACHOP 70
67 MACHOKE 5
69 BELLSPROUT 459
70 WEEPINBELL 23
71 VICTREEBEL 1
72 TENTACOOL 48
73 TENTACRUEL 4
74 GEODUDE 56
75 GRAVELER 3
77 PONYTA 1
79 SLOWPOKE 84
80 SLOWBRO 2
81 MAGNEMITE 11
82 MAGNETON 1
84 DODUO 1
86 SEEL 139
87 DEWGONG 5
88 GRIMER 10
90 SHELLDER 3,822
91 CLOYSTER 3
92 GASTLY 422
93 HAUNTER 19
94 GENGAR 4
95 ONIX 7
96 DROWZEE 1,317
97 HYPNO 33
98 KRABBY 554
99 KINGLER 22
100 VOLTORB 10
102 EXEGGCUTE 15
103 EXEGGUTOR 1
104 CUBONE 28
106 HITMONLEE 1
107 HITMONCHAN 3
108 LICKITUNG 8
109 KOFFING 59
110 WEEZING 1
111 RHYHORN 55
112 RHYDON 3
116 HORSEA 285
117 SEADRA 12
118 GOLDEEN 162
119 SEAKING 20
120 STARYU 128
121 STARMIE 1
123 SCYTHER 272
124 JYNX 14
125 ELECTABUZZ 37
126 MAGMAR 3
127 PINSIR 2
128 TAUROS 19
129 MAGIKARP 304
131 LAPRAS 3
133 EEVEE 2,099
134 VAPOREON 2
135 JOLTEON 1
140 KABUTO 1
143 SNORLAX 2
147 DRATINI 20
148 DRAGONAIR 1
149 DRAGONITE 3
Reading with the word “Pokémon” 54 times can induce semantic satiation.