Data Official Smogon University Usage Statistics Discussion Thread, mk.2

Status
Not open for further replies.

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
Sorry if this is a stupid question, but shouldn't the cutoff for dropping megas be lower than normal mons? You can only use one per team, and that's why their usage is so low compared to other pokemon.
A few things: precedent--you can only use one Rotom per team--and the logic behind the cutoff is still the same: tiers are lists of common Pokemon, so if a Pokemon isn't seen on at least 3.41% of teams, it's not COMMON--the reason why is immaterial.
 
"Real" win percentage for 1600 and up would be amazing still:

1. Good use of the real stat.
2. Better application of the max gxe thing.
3. Making it only 1600 (or whatever number) and up would mean we don't have new players messing up the pokemons' stats.
4. Useful statistic to cite when discussing potential suspects/bans and viability rankings. It would be like Chess where theoretically the best opening is e4 but statistics from Grandmaster battles show that d4 has the best win rate.
5. Could also be used to balance the levels of the pokemon in random battles.

I can't think of any reason not to make it. Well, besides time and work restraints obviously.
 

toshimelonhead

Honey Badger don't care.
is a Tiering Contributor
"Real" win percentage for 1600 and up would be amazing still:

1. Good use of the real stat.
2. Better application of the max gxe thing.
3. Making it only 1600 (or whatever number) and up would mean we don't have new players messing up the pokemons' stats.
4. Useful statistic to cite when discussing potential suspects/bans and viability rankings. It would be like Chess where theoretically the best opening is e4 but statistics from Grandmaster battles show that d4 has the best win rate.
5. Could also be used to balance the levels of the pokemon in random battles.

I can't think of any reason not to make it. Well, besides time and work restraints obviously.
Net win percentage is probably better, taking out games where both players use the same pokemon.
 
Does it take less time to generate stats if there's less things to have stats for? If so, what is exactly is the purpose in having usage stats for randomized formats when it can save time to gets stats up sooner for official tiers, and more specifically, tiers that are affected by fluctuations of usage, if stats for randomized tiers weren't accounted for?
 
Last edited:
Does it take less time to generate stats if there's less things to have stats for? If so, what is exactly is the purpose in having usage stats for randomized formats when it can save time to gets stats up sooner for official tiers, and more specifically, tiers that are affected by fluctuations of usage, if stats for randomized tiers weren't accounted for?
Secret trick of a randbats players: you can use the monthly stats as a cheat sheet to know what moves your opponent's pokemon might have. The sets do change occasionally so you want to use the most recent available stats, but they can reveal a lot of information that can help you make more informed decisions in the match. For instance, if you see a Golem who didn't have an Air Balloon and used Stealth Rock, you can be guaranteed he's carrying a Custap Berry. Notably Golem never runs weakness policy in randbats, so that's one threat you don't need to play around. It's a huge advantage to be able to know these things and make determinations about your opponent's sets based on that info.
 

Ununhexium

it's party time!
is a Contributor Alumnus
Secret trick of a randbats players: you can use the monthly stats as a cheat sheet to know what moves your opponent's pokemon might have. The sets do change occasionally so you want to use the most recent available stats, but they can reveal a lot of information that can help you make more informed decisions in the match. For instance, if you see a Golem who didn't have an Air Balloon and used Stealth Rock, you can be guaranteed he's carrying a Custap Berry. Notably Golem never runs weakness policy in randbats, so that's one threat you don't need to play around. It's a huge advantage to be able to know these things and make determinations about your opponent's sets based on that info.
Yeah but if you're playing randbats do you really want to go through that trouble to figure out its item
 
Yeah but if you're playing randbats do you really want to go through that trouble to figure out its item
You can use it as a guide to narrow down items, moves, and abilities. I just used Golem's item as an example. As for the trouble, it's incredibly easy. Just bookmark the most recent stats page once per month, then open it and control-F the pokemon you're looking for during the match. It's practically instant, and can help you easily rule out or confirm what kind of threat you're dealing with.
 
dodmen, I used to list Nature separately from EVs, but people seemed to find that LESS useful than the current configuration.

I've considered having a separate section for each stat, so it would look like:

Atk
252+: 10%
252: 6%
248: 1%
128: 4%
0: 58%
0-: 21%

Or something like that, but then you lose the correlation. I've also considered combining similar sets (e.g. 4/252/0/0/0/252 and 0/252/4/0/0/252) but there's a lot of strategy that goes into placing those last four EVs. The best I can recommend is that you look into the JSON files in the chaos folders, which contain the usage of EVERY spread and then maybe write your own analysis to combine them as you'd like. I don't know if you have any sort of programming background, but JSONs are super easy to work with, especially in languages like Python.
I am, for whatever reason, feeling inclined to mess around with that JSON. I was wondering what the numbers to the right of spreads/items/moves/checks and stuff exactly meant. Some are in the thousands and some are below one but greater than zero. Thanks.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
I am, for whatever reason, feeling inclined to mess around with that JSON. I was wondering what the numbers to the right of spreads/items/moves/checks and stuff exactly meant. Some are in the thousands and some are below one but greater than zero. Thanks.
Weighted count. Note that all weights are between zero and one.
 
Weighted count. Note that all weights are between zero and one.
I suppose I'm still not entirely sure what you mean by that. Swellow is shown closest to the top of the JSON and clocks in at around 17 for Scrappy and 104 for Guts. Look a little further and you see that Skyplate is sitting at .005. What am I supposed to take from these numbers? Thanks for the help.
 

Antar

is a Battle Server Administratoris a Programmeris a Super Moderatoris a Community Contributor
Official Data Miner
samiam541 Sum all the abilities. That gives you total count (I don't think I bothered to include a total count entry). So that would be 121 for Swellow.

Now use that as the denominator for any moveset stats you wish to compute. So:. 005/121 Swellows carry a Sky Plate.
 
Hi there Antar,

First off, thank you for the data you provide!

I'm looking to sort out Teammates in the moveset data set, and can't seem to figure it out.

I don't think that the .txt Teammates presented are simply the top teammates, as Kyogre-Primal in the VGC2016 data has no restricted legendaries in its top teammates. Therefor, I figure I must be reading the data set wrong.

I similarly have tried to make sense of the Teammate data in the JSON files, but don't really understand the Teammate values being output there either (some negative values?). Is there a calculation I'm missing, in order to get the correct top X teammates for a Pokemon? I've tried normalizing them over total abilities per mon, as it's been suggested for other calculations, but I don't believe that is providing correct data either, as Combee is coming in as Groudon-Primal's top partner haha.

Please let me know if there is any insight you can shed to help me move forward!

Thank you in advance!
 

toshimelonhead

Honey Badger don't care.
is a Tiering Contributor
Hi there Antar,

First off, thank you for the data you provide!

I'm looking to sort out Teammates in the moveset data set, and can't seem to figure it out.

I don't think that the .txt Teammates presented are simply the top teammates, as Kyogre-Primal in the VGC2016 data has no restricted legendaries in its top teammates. Therefor, I figure I must be reading the data set wrong.

I similarly have tried to make sense of the Teammate data in the JSON files, but don't really understand the Teammate values being output there either (some negative values?). Is there a calculation I'm missing, in order to get the correct top X teammates for a Pokemon? I've tried normalizing them over total abilities per mon, as it's been suggested for other calculations, but I don't believe that is providing correct data either, as Combee is coming in as Groudon-Primal's top partner haha.

Please let me know if there is any insight you can shed to help me move forward!

Thank you in advance!
The teammate data is comparing usage of mon with teammate to the teammate's normal usage. That's where the negative numbers come from.
 
Status
Not open for further replies.

Users Who Are Viewing This Thread (Users: 1, Guests: 0)

Top