This is a topic that has always interested me, and today i quite randomly found out that shoddybattle.com provides an easy to use, n-pokémon usage grapher for every pokémon! Not only can you plot the usage of any mon you want, but you can, more interestingly compare usages between as many pokémon as you like!
This is basically an explicit representation of metagame trends, and is quite exciting if you know what to look for. You'll need to make assumptions based on basic binary microinstances in battles (or how pokemon face off against eachother in general). Obviously one can never truely express all the cause and effect variables involved, but there are some pretty nice relationships. Here's some examples:
Celebi Vs. Breloom
Celebi is as close to a paper counter for breloom as one can hope for; with dual stab resistance, natural cure, physical bulk and recover, there is really no chance of breaking any healthy celebi with any breloom set, no matter how specialized. Taking this into consideration i predicted that one's usage might be inversely proportional to Celebi's, and was surprised with this lovely dataset:
this not only displays a pretty explicit relationship, but also shows that there's always more at play than a simple binary feedback: note the low starting point for celebi (probably due to ttar/weavile/gengar's gen 4 boosts) and the high starting point for loom (who gained poison heal and grass stab). Also note the plunge in celebi's usage post plat (probably due to scizor)
Blissey Vs. Zapdos
Zapdos can literally do nothing to blissey exept possibly toxic stall it with the subroost set. Blissey on the other hand can status it and stoss/ib zappy to death. with this, one would predict a negative growth in zapdos' usage with a positive growth in blisseys usage... sure enough:
note that there is possibly some relationship here (slow decline in zapdos usage with a slow increase in blissey usage) but there's much more at play here and the relationship isn't anywhere near as "obvious" as the celebi-breloom one. Note also the incrase in zapdos usage with the addition of heat wave in Pt and the decrease in plissey usage with the prevalence of trick.
Abomasnow Vs. Walrein
This one seems obvious, esp with X-act's new "likely teammates" feature putting these guys at 90%(!) likely to be on a team together. KNowing this, we expect a correlated set of usage variations...
Ha, an almost one to one correlation. not much to say here, although hail teams on the whole seem to be stable on the official server, with a surprising slight increase post Pt, despite scizor's awesomeness!
Salamence Vs. Magnezone Vs. Bronzong
lastly we have a 3 dimensional analysis. the idea was that bronzong stops most salamences, and magnezone dispatches most bronzongs, as well as forming good defensive synergy with mence. with this in mind, one would predict a proportional relationship between mence and zone, with an inverse proportional relationship between zong and mence/zone. What we get is....
Evidence that predictions get trickier with three variables! It's been pretty tough for me to isolate anything besides binary "relationships". hopefully soemone else will have more success than me itt!
Celebi Vs. Heatran Vs. Infernape
The famous "CeleTran" combination is a great defensive core, but has difficulties with almost all infernape varieties, who's stabs OHKO both of them. With this in mind, we expect Celebi's usage to grow with Heatran's, and both to decrease with Infernape usage.
Here we have a nicer display of a predictable relationship between Bi and Tran, and a bit of a noisy signal form Nape, implying once again that it's tough to isolate a three way relationship.. if anything this data seems to imply that nape's usage has been moderated by celetrans growth which is obviously nonesense!
__________
Obviously there are problems with assuming that one pokemon doing well against the other will result in a usage relationship every time. Gliscor and heracross for example display little effect on eachother over the months, and machamp and tyranitar seems frustratingly unrelated as well.
The only real downside is that it uses shoddy stats and not smogon server stats. it would be fucking awesome if we had own own simple trend grapher but in the meantime this one has plenty of interesting tidbits. We could even link usage increases with posted RMTs or pokemon centred threads (e.g. Rhyperior), but it's not like our programmers aren't already bogged down with work Ω__Ω. I'm confident that the usage trends would be even nicer with smogon data though!
so, discuss, and present your own analyses of trends!
This is basically an explicit representation of metagame trends, and is quite exciting if you know what to look for. You'll need to make assumptions based on basic binary microinstances in battles (or how pokemon face off against eachother in general). Obviously one can never truely express all the cause and effect variables involved, but there are some pretty nice relationships. Here's some examples:
Celebi Vs. Breloom
Celebi is as close to a paper counter for breloom as one can hope for; with dual stab resistance, natural cure, physical bulk and recover, there is really no chance of breaking any healthy celebi with any breloom set, no matter how specialized. Taking this into consideration i predicted that one's usage might be inversely proportional to Celebi's, and was surprised with this lovely dataset:
this not only displays a pretty explicit relationship, but also shows that there's always more at play than a simple binary feedback: note the low starting point for celebi (probably due to ttar/weavile/gengar's gen 4 boosts) and the high starting point for loom (who gained poison heal and grass stab). Also note the plunge in celebi's usage post plat (probably due to scizor)
Blissey Vs. Zapdos
Zapdos can literally do nothing to blissey exept possibly toxic stall it with the subroost set. Blissey on the other hand can status it and stoss/ib zappy to death. with this, one would predict a negative growth in zapdos' usage with a positive growth in blisseys usage... sure enough:
note that there is possibly some relationship here (slow decline in zapdos usage with a slow increase in blissey usage) but there's much more at play here and the relationship isn't anywhere near as "obvious" as the celebi-breloom one. Note also the incrase in zapdos usage with the addition of heat wave in Pt and the decrease in plissey usage with the prevalence of trick.
Abomasnow Vs. Walrein
This one seems obvious, esp with X-act's new "likely teammates" feature putting these guys at 90%(!) likely to be on a team together. KNowing this, we expect a correlated set of usage variations...
Ha, an almost one to one correlation. not much to say here, although hail teams on the whole seem to be stable on the official server, with a surprising slight increase post Pt, despite scizor's awesomeness!
Salamence Vs. Magnezone Vs. Bronzong
lastly we have a 3 dimensional analysis. the idea was that bronzong stops most salamences, and magnezone dispatches most bronzongs, as well as forming good defensive synergy with mence. with this in mind, one would predict a proportional relationship between mence and zone, with an inverse proportional relationship between zong and mence/zone. What we get is....
Evidence that predictions get trickier with three variables! It's been pretty tough for me to isolate anything besides binary "relationships". hopefully soemone else will have more success than me itt!
Celebi Vs. Heatran Vs. Infernape
The famous "CeleTran" combination is a great defensive core, but has difficulties with almost all infernape varieties, who's stabs OHKO both of them. With this in mind, we expect Celebi's usage to grow with Heatran's, and both to decrease with Infernape usage.
Here we have a nicer display of a predictable relationship between Bi and Tran, and a bit of a noisy signal form Nape, implying once again that it's tough to isolate a three way relationship.. if anything this data seems to imply that nape's usage has been moderated by celetrans growth which is obviously nonesense!
__________
Obviously there are problems with assuming that one pokemon doing well against the other will result in a usage relationship every time. Gliscor and heracross for example display little effect on eachother over the months, and machamp and tyranitar seems frustratingly unrelated as well.
The only real downside is that it uses shoddy stats and not smogon server stats. it would be fucking awesome if we had own own simple trend grapher but in the meantime this one has plenty of interesting tidbits. We could even link usage increases with posted RMTs or pokemon centred threads (e.g. Rhyperior), but it's not like our programmers aren't already bogged down with work Ω__Ω. I'm confident that the usage trends would be even nicer with smogon data though!
so, discuss, and present your own analyses of trends!