Data Visualising metagame trends

[Imran]

Previously, usage data has been analysed on a month-by-month basis, as and when new statistics become available. This is great for looking at short term changes, for example the impact of a particular centralising Pokemon being banned from a tier. But what about longer term trends? You need a visual tool. Fortunately, such a thing now exists and can be found at the URL below:

http://pokestat.org.uk/

I haven't played properly since DPP, so I'm not the best person to use this tool. I'm very interested to see what people who are more involved in the game as it stands today are able to come up with! Please use and abuse this tool to find out interesting things and post them back in this thread! If you want some ideas, some of my exploration is in the spoiler section below:

Imran's Analysis

Let's take a Pokemon I used to use a lot back in the day - Gengar. Add Gengar to the graph - note how its usage has steadily increased since about April. What could be causing this? What's a great Pokemon to switch into Gengar? When I played Blissey was an obvious choice, add Blissey to the graph too - note how its usage has been steadily decreaing since about April. This may explain why Gengar has been growing in popularity over that period! Quite what's causing this steady decrease in Blissey usage isn't entirely clear to me though. Could it be the steady increase in the use of Keldeo since February?

Clear the graph, put Skarmory and Garchomp on there. Note how as one's uasge increases the other decreases. These numbers fluctuate quite rapidly. Quite what causes them to fluctuate so much I don't know. But we can see from the graphs that their usage number are quite strongly inversely correlated!

Again, I'm not familiar with XY, so please go ahead explore those ideas and your own. Let us know what you find out! I wish I could post this in a more general thread, as it covers more than just OU, but this seems the most appropriate place that I have permission to post in.

For anyone interested in the technical details, here is the Github repository. I implemented the backend in the Play Framework in Scala, and the front end in Bootstrap, jQuery and Chart.js. Please feel free to raise issues there if you find any bugs, reporting them in this thread will also work. You may find that this doesn't work well on mobile devices or on machines with low-resolution - my front-end development skills aren't exactly stellar.

Happy stat hunting!

 

[Arcticblast]

Doubles seems to be broken. This is damn cool though, props.

 

[firehusky]

Wow this is really cool!
Wish it had the ORAS stats though
Something I found out for OU was when the usage increased for greninja, the usage for azumarill went up by a lot, since it was one of the best counters for greninja, other than empoleon and chansey, both of which have gotten pretty rare. (For some reason, I don't see chansey a lot, maybe because stall isn't so popular anymore) And then when the usage for azumarill went up, ferrothorn's usage went up too.
So, because greninja's usage went up, ferrothorn's usage went up!
EDIT: I put in heatran, and it's usage also went up by quite a bit. Next, I put in lando-t, which usage also went up by a lot. Scarf landog checks heatran pretty well, and greninja.
Therefore greninja is overcentralizing in this metagame and should be banned
This is really interesting :]

 

[Wumgum]

And then when the usage for azumarill went up, ferrothorn's usage went up too.
So, because Aegislash's usage went down, ferrothorn's usage went up!

fixed

Hide

 

[Burt Racecar]

Wasn't originally planning on posting but I noticed a little trend between some of the ORAS megas. When they were announced their base form usage spiked and then fell away again. All this is OU data by the way.

-Salamence back in January being on 3.6% of teams at it's highest and then fell to 3% by June, before rising coincidentally to spike at 3.3% in August , the month when it was announced.
-Sableye shows a similar story starting high being on around 4.5% of teams, and by May fell to just above 3% before spiking up to 3.6% at the time its Mega form was announced.
-Sceptile sat on an abysmal 0.6% of teams before peaking to 0.9% of teams at the time its Mega was announced, it shows a few were psyched enough by his Mega form to put a place for him on his team for a short time.
-Swampert's graph is a funny looking one, starting on 1.4% of teams in January spiking at feb/mar at 1.6%, but something pooped on it and it fell to 1.5% by May. All of a sudden its usage shot up in June to 1.7%, again at the time its Mega form was announced.
-Metagross was found being on 2.2% of teams in Jan and then decreased steadily to 1.9% by June, then its Mega form was announced in mid July and it surpassed its previous peak and reached 2.3%, before plummeting at the same rate as its increase.
-Gallade's wobbly line, rising and falling between 0.5% and 0.6% is too hard to analyse based on the announcement of its mega form, though its usage did happen to increase at that particular time.
-Slowbro's graph is the only one that maintained a high usage after its mega form was announced, and indeed its usage grew exponentially from August to October, effectively doubling from being on 3% of teams to finding itself on 6% of teams. However this was probably due to the Aegislash ban, but I personally like to think it was really because people were hyped beyond belief for shell armor Slowbro!

Couldn't do it for all Megas because their OU usage wasn't high enough.

This is a really neat tool, thanks a lot!

 

[firehusky]

fixed

Hide

Don't really see what that has to do with anything.. aegislash was banned, and I do not see the relationship between both of those mons

 

[Imran]

Doubles seems to be broken. This is damn cool though, props.

Hey yeah, as discussed, this is an issue that's been noted on Safari on iOS. Damn Apple for making it so hard to debug, apparently I need a mac to do that - or to go back to iOS 7 :( Thanks for nothing this - I've raised an issue.

Analysis + ORAS

Really interesting to note the joint rise of Azumarill, Ferrothorn and Heatran. Their graphs are almost exactly the same! You seem to be suggesting that it's because they counter each other quite well... I'm wondering if there's some good pairing example that can be discovered - Pokemon with good synergy. I remember people raving about SkarmBliss in RS (long time ago!) I bet those two went up and down together.

As for ORAS - stats are just starting to come in. I hope to start basing some of this stuff of the data collected here to hopefully cover more generations and tiers - but as a lot of it's been lost it will take some time to get good long-term data that will make a chart like this worth while! I'm currently using the data from some of the sources where it's been backed up - but it's a tricky job finding it! If anyone happened to keep any of the old files I'd be extremely happy!

Mega Analysis

This is a great piece of analysis that I hadn't though of, you can genuinely watch the hype build and fail! The Slowbro one is especially interesting. I trust what you have to say about Aegislash, I wonder if the fact that the Mega hype made people use slowbro also means it was "discovered", as in people hadn't thought of using it before until the Mega announcement, and then realised it was actually viable. Thanks for the insight :)

 

[Imran]

fixed

Hide

Don't really see what that has to do with anything.. aegislash was banned, and I do not see the relationship between both of those mons

All three seems to start rising more steadliy after the Aegislash ban - is it possible that they all fit into certain styles of team better now that it's gone?

 

[Justintown11]

Wow Imran, thanks for the awesome resource.
Btw, has any noticed the usage for landorus-therian? In 5 months it's skyrocketed from 7.5% to nearly 20%

 

[boltsandbombers]

Wow Imran, thanks for the awesome resource.
Btw, has any noticed the usage for landorus-therian? In 5 months it's skyrocketed from 7.5% to nearly 20%

Yeah, I think it's the most common Pokemon in OU. Also, I could have sworn it was around 25% at some point.

 

[MilkyWay]

Wow Imran, thanks for the awesome resource.
Btw, has any noticed the usage for landorus-therian? In 5 months it's skyrocketed from 7.5% to nearly 20%

Yeah, I just saw that too. I think thats around the time when the scarf set became really popular

 

[Wheezer]

The graphs of Tyranitar and Excadrill are also similar. Both of these mons were used a good amount, but not as much at the beginning of X/Y, because of all the powerful threats that had yet to be banned. The popularity of Sand Offense sprouted up in May and June, and the graphs of Tyranitar and Excadrill emphasize this. There is a huge jump from April into May. I am not quite sure of why this is, but I assume it is because of some pokemon getting banned (I am drawing a blank right now and can't think). The graphs also show how the usage of Sand Offense died down after that.

I believe that as we look at more and more at different cores, we will be able to see that the usage stats of pokemon reflect metagame shifts and changes that have happened throughout X/Y, and in other generations too.

 

[Consignia]

I'm inclined to think that Sand Rush Excadrill was held back by the ludicrously high Rotom-W usage in early X/Y. When Rotom-W began to fall off, Sand Rush Excadrill was able to shine- the surge in Excadrill happens just when Rotom-W usage has stabilized to the level it's still at today.

Hide

One Pokemon that I think is interesting to look at is Breloom. Breloom wasn't rated very highly at the start of X/Y, partly because of the new sleep mechanics and Low Sweep nerf, partly because it struggled to handle many of the new threats (cough cough, Mega Venusaur). Breloom usage clearly rises and falls in response to the prevalence of Excadrill and Tyranitar. I feel this is interesting since Breloom isn't really a sand pokemon itself, but it's a fantastic response to sand: sand teams can't really run Mega Venusaur and it wasn't uncommon to see them forgo grass types entirely, so their best answer to Loomer was often a lati twin or something similar. Breloom revenge killed Excadrill, bullied Tyranitar, and its coverage plus Spore gave it an excellent matchup vs. common sand teams at the time. Therefore, the delay on the curve might give us a rough estimate of the time it takes the metagame to adapt and respond to a trend, in this case, the Sand Rush Excadrill trend.

Thanks for providing this tool, it's quite fun to play around with.

 

[BlueLobster]

Landorus' forms are strongly linked. Before Lando-T's spike in usage, Lando-I was at it's highest usage point. It is noted this was one of Lando-T's lowest points.

Gliscor usage dropped a small amount at Lando-T's time of dominance, but not enough to call it noteworthy.

Also, Chansey/Conkeldurr usage is highly related, as you'd expect.

Chansey usage had risen starting at June, and peaked around 2 months later. So I added Conk to the graph, and his usage had sloped down from then on. Both evened out around the same time.

 

[boltsandbombers]

Chansey/Conkeldurr usage is highly related, as you'd expect.

Chansey usage had risen starting at June, and peaked around 2 months later. So I added Conk to the graph, and his usage had sloped down from then on. Both evened out around the same time.

Please do not double post, use the edit function.

 

[BlueLobster]

Please do not double post, use the edit function.

Yeah sorry about that, my phone glitched and I couldn't see my old post.

@Below: Other banned mons and stuff overhyped at the begining of the XY meta (such as Carbink for some reason) have negative usage on the graph. I'm guessing the lines are slighty misplaced.

 

[Zebstrika]

So Mawile actually had negative usage in september. If you trust the graph.

I was actually seeing if its ban caused its niche counters like Arcanine and Weezing to drop. Sadly, they didn't.

 

[Wheezer]

I'm inclined to think that Sand Rush Excadrill was held back by the ludicrously high Rotom-W usage in early X/Y. When Rotom-W began to fall off, Sand Rush Excadrill was able to shine- the surge in Excadrill happens just when Rotom-W usage has stabilized to the level it's still at today.

Hide

One Pokemon that I think is interesting to look at is Breloom. Breloom wasn't rated very highly at the start of X/Y, partly because of the new sleep mechanics and Low Sweep nerf, partly because it struggled to handle many of the new threats (cough cough, Mega Venusaur). Breloom usage clearly rises and falls in response to the prevalence of Excadrill and Tyranitar. I feel this is interesting since Breloom isn't really a sand pokemon itself, but it's a fantastic response to sand: sand teams can't really run Mega Venusaur and it wasn't uncommon to see them forgo grass types entirely, so their best answer to Loomer was often a lati twin or something similar. Breloom revenge killed Excadrill, bullied Tyranitar, and its coverage plus Spore gave it an excellent matchup vs. common sand teams at the time. Therefore, the delay on the curve might give us a rough estimate of the time it takes the metagame to adapt and respond to a trend, in this case, the Sand Rush Excadrill trend.

Thanks for providing this tool, it's quite fun to play around with.

Ya, I see what you mean about Rotom-W repressing the usage of Excadrill. Although, Rotom-W usage has king of leveled off for the past few months, but the usage of Sand Offense is still going down, so that's a little wierd. It is also a bit eerie how close the usage stats of Breloom are to Sand Offense. You would think that the usage of other mons that check Sand Offense would be going down too, like Breloom, but if you look at the usage of mons like Keldeo and Conkeldurr, they kind of have opposite trends of Sand Offense usage. Keldeo was rising up in usage after the peak of Sand Offense, but Sand Offense went down. Conkeldurr has a bit of a similar graph to Sand Offense, but instead of peaking where Sand Offense did, it fell.

 

[Outrageous Fortune]

Fantastic work imran, this is incredibly fun to play around with. One thing though; would it be possible to restrict the shown usage to certain months so that it's possible to see the smaller fluctuations more easily? Either this or a zoom tool. It's a very minor problem though and only affects things with a high range of usage (eg. landorus-t).
Again, thanks a heap for making this.

 

[Imran]

The graphs of Tyranitar and Excadrill are also similar...

Above +Rotom-W + Breloom

Above + Keldeo

I plotted all 5 of the Pokemon mentioned above on to it. Really interesting to note the "discovery" of Sand Offence between April and May - I'd like a bit of insight here, though. Is that a popularity trend, or is it based off other factors? You noted that it spiked as Rotom-W usage hit around where it's at today - are these related in some way? I can't think how that would be the case at the moment, but these factors are always hidden. There's probably some other Pokemon / set of Pokemon we can plot on here to perfectly explain this!

Landorus' forms are strongly linked. Before Lando-T's spike in usage, Lando-I was at it's highest usage point. It is noted this was one of Lando-T's lowest points.

Gliscor usage dropped a small amount at Lando-T's time of dominance, but not enough to call it noteworthy.

Also, Chansey/Conkeldurr usage is highly related, as you'd expect.

Chansey usage had risen starting at June, and peaked around 2 months later. So I added Conk to the graph, and his usage had sloped down from then on. Both evened out around the same time.

Both very neat demonstrations of simple inverse relationships!

So Mawile actually had negative usage in september. If you trust the graph.

I was actually seeing if its ban caused its niche counters like Arcanine and Weezing to drop. Sadly, they didn't.

Interesting to note that they are slowly starting to drop - that'll be one to watch over the next two three months. If you project their current graphs linearly we will see a significant drop as time goes by. I suppose people are just trying to find out if they have value beyond simply countering Mawile. We'll have to see.

As for the Mawile plot - this is due to the way the graph is rendered. It uses something called a Bezier curve to interpolate between values, to make the graph look smoother and nicer - rather than just having each point joined up by straight lines which would make it look quite jagged. Note how actually on the actual data point for September and October the usage is at exactly zero. In 99% of cases this is harmless and makes the graph look a lot nicer, but it does make it look rather odd in this particular case!

 

[Grey Waves]

Will this application ever include support for RU or NU?

 

[firehusky]

Will this application ever include support for RU or NU?

I'd assume that they are probably working on it currently.

 

[Imran]

Fantastic work imran, this is incredibly fun to play around with. One thing though; would it be possible to restrict the shown usage to certain months so that it's possible to see the smaller fluctuations more easily? Either this or a zoom tool. It's a very minor problem though and only affects things with a high range of usage (eg. landorus-t).
Again, thanks a heap for making this.

Yes - potentially. You'd want to be able to specify the range of months? It's certainly possible, I've raised an enhancement request.

Will this application ever include support for RU or NU?

I hope so. If the old data hadn't all be lost I would indeed be working on it right now, but alas it's not. We have usage data from November onwards, and that covers all tiers and generations. I hope, in a few months - when we have data over a longer period again - to be able to cover all generations mentioned here. In the meantime, while we only have data for a single month, I'm going to be working on shorter term visualisations.

 

[Jaroda]

Something important would be to see differences in mega and non mega forms. Or something like Charizard with two completely different viable forms.

 

[silver97]

Its quite impressive to see how aegi's ban affected the meta, there are a shitload of mons that have seen more usage after its ban, including medicham, gardevoir, heracross, pinsir, slowbro, jirachi, mew, starmie, even metagross had a spike of usage in that period and then fell down again.