GamePress

Wild Pokemon Movesets are Uniformly Distributed

Featured Contents

Introduction

Now that GamePress has a large dataset to work with, we can finally address some of the questions that long lingered on our minds. Our question today is:

Do Pokemon get some movesets more often than others?

To answer this, we sifted through our catch attempt database once again and sorted each Pokemon by their movesets. We plotted the starting movesets of all species of Pokemon that were encountered at least 100 times (see below) to visualize our data. We also ran a chi-square test for every sighted Pokemon, to check for any statistically significant deviations in their moveset distributions.

Our analysis suggests that all possible movesets are equally likely for wild pokemon.

Pokemon Moveset Distribution

These are pie charts of every Pokemon that was encountered at least 100 times:

Pokemon GO Moveset Distribution

Chi-Square Test

Our hypothesis was that for each Pokemon, all of their non-legacy movesets are equally likely to occur.

To check for statistical significance, we ran a chi-square test for every sighted Pokemon. This test looks at the the amount of times a moveset occurred in our sample and checks if that amount is statistically unlikely.

# Pokemon Encounters p-value
1 Bulbasaur 305 0.295213
2 Ivysaur 1 0.525678
4 Charmander 106 0.576888
5 Charmeleon 70 0.210231
6 Charizard 16 0.623388
7 Squirtle 859 0.037688
10 Caterpie 107 0.208842
11 Metapod 2 1.000000
12 Butterfree 3 0.913070
13 Weedle 139 0.552691
14 Kakuna 11 0.131668
16 Pidgey 4299 0.320530
17 Pidgeotto 29 0.154345
18 Pidgeot 6 0.849145
19 Rattata 101 0.941217
20 Raticate 1 0.525678
21 Spearow 107 0.688460
23 Ekans 53 0.073647
25 Pikachu 604 0.537089
26 Raichu 25 0.543671
27 Sandshrew 393 0.339479
28 Sandslash 7 0.938802
29 Nidoran ♀ 561 0.901379
30 Nidorina 368 0.869775
31 Nidoqueen 39 0.362182
32 Nidoran ♂ 1039 0.101412
33 Nidorino 363 0.335390
34 Nidoking 46 0.289662
35 Clefairy 744 0.695012
36 Clefable 153 0.042158
37 Vulpix 2 0.751212
39 Jigglypuff 177 0.273109
40 Wigglytuff 124 0.503966
41 Zubat 205 0.582874
43 Oddish 37 0.201303
44 Gloom 4 0.699986
46 Paras 511 0.542394
47 Parasect 206 0.396072
48 Venonat 624 0.889867
49 Venomoth 309 0.316750
50 Diglett 122 0.622148
51 Dugtrio 166 0.210969
52 Meowth 14 0.921163
54 Psyduck 2275 0.943904
55 Golduck 4 0.984748
56 Mankey 1 0.525678
58 Growlithe 196 0.134144
59 Arcanine 228 0.055718
60 Poliwag 891 0.825878
61 Poliwhirl 450 0.767433
62 Poliwrath 83 0.108107
63 Abra 128 0.681941
64 Kadabra 121 0.831740
65 Alakazam 14 0.335020
66 Machop 153 0.812867
67 Machoke 8 0.751212
68 Machamp 22 0.849145
69 Bellsprout 473 0.742268
70 Weepinbell 316 0.643587
71 Victreebel 40 0.578555
72 Tentacool 183 0.090984
73 Tentacruel 5 0.653842
74 Geodude 250 0.224839
75 Graveler 12 0.849145
76 Golem 31 0.601546
77 Ponyta 3 0.913070
79 Slowpoke 353 0.777135
80 Slowbro 183 0.570848
81 Magnemite 383 0.065062
82 Magneton 448 0.017979
84 Doduo 3 0.913070
85 Dodrio 3 0.913070
86 Seel 4 0.699986
88 Grimer 721 0.570975
89 Muk 18 0.376568
90 Shellder 159 0.877053
91 Cloyster 49 0.189332
92 Gastly 198 0.188970
93 Haunter 479 0.967458
94 Gengar 19 0.636700
95 Onix 1 0.525678
96 Drowzee 4 0.699986
98 Krabby 529 0.650573
102 Exeggcute 200 0.318224
103 Exeggutor 307 0.025745
104 Cubone 108 0.581833
105 Marowak 59 0.220640
108 Lickitung 3 0.415880
109 Koffing 6 0.606531
111 Rhyhorn 6 0.962566
114 Tangela 120 0.102797
116 Horsea 467 0.503273
117 Seadra 1 0.525678
118 Goldeen 76 0.104939
119 Seaking 7 0.026966
120 Staryu 884 0.575093
121 Starmie 170 0.676420
126 Magmar 1 0.525678
127 Pinsir 14 0.677972
128 Tauros 24 0.549416
129 Magikarp 3471 1.000000
131 Lapras 62 0.954389
133 Eevee 1119 0.994249
134 Vaporeon 325 0.552199
136 Flareon 1 0.513417
137 Porygon 1 0.525678
138 Omanyte 432 0.780673
139 Omastar 20 0.269219
140 Kabuto 525 0.111831
141 Kabutops 23 0.426585
143 Snorlax 404 0.946950
147 Dratini 764 0.934202
148 Dragonair 372 0.844209
149 Dragonite 82 0.741265

The vast majority of the dataset has a p-value above 0.05, which is not statistically significant to reject our hypothesis. While there are five cases with p < 0.05, this is to be expected when running > 100 independent tests. In addition, it is highly unlikely that Niantic would choose five seemingly arbitrary Pokemon to have non-uniform movesets. Considering this and our large sample size for many different species, we believe that for each species of Pokemon, movesets are equally distributed.