Thank you! So that's why!
-
Welcome to Smogon! Take a moment to read the Introduction to Smogon for a run-down on everything Smogon, and make sure you take some time to read the global rules.
Resource VGC: Ask a Simple Question, Get a Simple Answer
- Thread starter blarajan
- Start date
How to beat rain in Reg I?
My team for reference: https://pokepast.es/c9fdd7e99ca33e2b
I have been struggling in the match up against kyogre rain teams, especially power item kyogre + ice rider, and kyogre rain boosting urshifu rapid strikes.
I have torkoal to set my own weather condition, but I just can't win the weather war in many battles, especially when torkoal instantly loses if rain is set while it is on field.
If anyone has any strategies or any edits to my team I could make to deal with this that would be super helpful! I am considering dropping smeargle, but I don't know what I would replace it with.
My team for reference: https://pokepast.es/c9fdd7e99ca33e2b
I have been struggling in the match up against kyogre rain teams, especially power item kyogre + ice rider, and kyogre rain boosting urshifu rapid strikes.
I have torkoal to set my own weather condition, but I just can't win the weather war in many battles, especially when torkoal instantly loses if rain is set while it is on field.
If anyone has any strategies or any edits to my team I could make to deal with this that would be super helpful! I am considering dropping smeargle, but I don't know what I would replace it with.
Hello,
I'm not sure if this is the right place to post this, so I apologize in advance if it is not.
I'm currently trying to create a legally distinct program similar to the current VGC format for a school project. In the game, the player will battle a simple AI I've created, with a few preset teams that are inspired by popular and strong teams in VGC Reg I. The program will be simplified greatly compared to actual Pokemon, with types at the moment, less moves and one less status effect (those being included in the game are burn, para, poison and sleep). The player will have access to a few different teams, all of which will be modeled after the current meta teams. The team that the AI uses will be chosen from a random set of preset teams. to try to reduce repetition, but there is only so much I can do, especially if the player brings the same team every time.
The pokepastes of some of the teams I plan to add are:
Choice specs Miraidon and clear amulet Groudon team:
https://pokepast.es/313e35a77bf4c61e
Life orb Calyrex-Shadow and Zamazenta-C team:
https://pokepast.es/89d18c940f39794f
Clear amulet Calyrex-Ice and power herb Lunala trick room team:
https://pokepast.es/3191de3d1789cbff
(Of course everything will have different names and such).
If anyone has any suggestions for changes to any of these teams or other teams you'd like me to add, please tell me.
I don't plan on adding tera - I feel like it will be a nightmare to code - but if you think it is a necessary thing, please reply to tell me that, and I will attempt to add it.
As for the AI, I plan to give it simple prediction by providing it access to every damage calc for every mon on the field, and if an outspeed 2HKO or OHKO is threatening any of its mons, it will switch (unless it can't switch). This is obviously limited in terms of its ability to understand which mon it is switching the threatened mon out for and overall why it is bringing that other mon in. If there is less of a threat of an outspeed 2HKO or OHKO, then the AI will either choose to setup or attack. If anyone has any ideas how I can improve this while keeping it simple, please let me know.
The aim of this program is to be a simplified, strategic and fun version of VGC that allows completely new players to learn some
Finally, I'd like to know if anyone has any suggestions at all for other things I should add or may have missed, or any changes you think would make it a more strategic and enjoyable game.
Thanks for reading all this, I hope I didn't waste your time and have a good day!
I'm not sure if this is the right place to post this, so I apologize in advance if it is not.
I'm currently trying to create a legally distinct program similar to the current VGC format for a school project. In the game, the player will battle a simple AI I've created, with a few preset teams that are inspired by popular and strong teams in VGC Reg I. The program will be simplified greatly compared to actual Pokemon, with types at the moment, less moves and one less status effect (those being included in the game are burn, para, poison and sleep). The player will have access to a few different teams, all of which will be modeled after the current meta teams. The team that the AI uses will be chosen from a random set of preset teams. to try to reduce repetition, but there is only so much I can do, especially if the player brings the same team every time.
The pokepastes of some of the teams I plan to add are:
Choice specs Miraidon and clear amulet Groudon team:
https://pokepast.es/313e35a77bf4c61e
Life orb Calyrex-Shadow and Zamazenta-C team:
https://pokepast.es/89d18c940f39794f
Clear amulet Calyrex-Ice and power herb Lunala trick room team:
https://pokepast.es/3191de3d1789cbff
(Of course everything will have different names and such).
If anyone has any suggestions for changes to any of these teams or other teams you'd like me to add, please tell me.
I don't plan on adding tera - I feel like it will be a nightmare to code - but if you think it is a necessary thing, please reply to tell me that, and I will attempt to add it.
As for the AI, I plan to give it simple prediction by providing it access to every damage calc for every mon on the field, and if an outspeed 2HKO or OHKO is threatening any of its mons, it will switch (unless it can't switch). This is obviously limited in terms of its ability to understand which mon it is switching the threatened mon out for and overall why it is bringing that other mon in. If there is less of a threat of an outspeed 2HKO or OHKO, then the AI will either choose to setup or attack. If anyone has any ideas how I can improve this while keeping it simple, please let me know.
The aim of this program is to be a simplified, strategic and fun version of VGC that allows completely new players to learn some
Finally, I'd like to know if anyone has any suggestions at all for other things I should add or may have missed, or any changes you think would make it a more strategic and enjoyable game.
Thanks for reading all this, I hope I didn't waste your time and have a good day!
Last edited:
This is actually a really cool idea! A few things I have to note:
I think the first (and most obvious) thing that comes to mind here as a avid VGC enjoyed is that VGC seems very difficult for AI to actually comprehend at the moment. So many things could happen and the fact that it has to take in so much information may make it difficult for the AI to handle. I think you are doing the right thing here though (removing Tera, freeze, and giving it information) to make the AI actually work.
Of course you would also need to find an actual reasoning for this to be used over something like Showdown. While yes, I understand it is for newer players, but it could be far easier for a new player to watch a free video online on teambuilding / battle analysis / whatever and hop on ladder and be fine. If you push this out enough this would probably be fine though.
I do have a few questions that I'll list out quickly.
- Would this be run under Open Team Sheets or Closed Team Sheets
I think it would be best to run this under both, giving the player the option to either learn how to take in information and use it to their advantage or giving them the option to try and figure out how to obtain said information and abuse it. This would also raise the question; would it be best of 1 battles or best of 3? I think it would probably be best to replicate how the ladder does it, especially since best of 1's in the competitive VGC scene are limited to online, CTS tournaments (and some very limited regional tournaments) + VGC team tournaments held here on Smogon.
I see in your post you outline 3 teams that you intend to add to this simulator. Are those all the teams you plan to add or will more be implemented? I think this thing shouldn't be up and running until you cover all the big restricted cores right now, like CSR+ZamaC and Mirai+Icerider. So maybe about 10 teams for a initial launch?
Keeping this AI and simulation up to date just on your own could also be pretty difficult, considering competitive Pokemon as a whole (not just VGC) is a ever changing format. Like literally we rotate formats LOL.
Nope! That's enough yap from me (the ethics of this dont really matter as long as the AI isn't building teams for the player). Really cool idea, hope it works out! I think here would also be a nice place to ask feedback (don't forget to read the rules of said subform!). Good luck trainer!
I think you misunderstand the project here (and I can be corrected if I’m wrong). In this case I don’t think AI is being used to describe some algorithm trained on statistics or anything, but a more simple algorithm with some more simple instructions that can mimic a human opponent. There aren’t ethical considerations for this, unless you really hate IKEA or something.
As for the project, I’d recommend reading up a little bit on game theory and nash equilibrium to help build something like this. Seems like a fun project, it would be cool to see it when it’s finished!
Hi, thanks for your amazing reply!
I agree that the AI will likely be extremely limited in its capacity to understand VGC, and therefore limited in its capacity to provide a challenge to the player. I am exploring different ideas to minimize this limitation, but I'd like to do that without creating artificial difficulty, for example just giving the AI higher level mons. I want to keep the battle as fair as possible while still making it challenging.
I am struggling to come up with a problem that this program solves to be honest. With the presence of Showdown, a much more complex and accurate simulator, I need something different; I think the AI provides this and, after reading your opinion on this, I think it would be better for me to shift the focus of this program from the player to the AI; there are a lot of resources for learning VGC (I myself have only been playing for two or so months), but there is less of an exploration of AI. Maybe exploring how different AI strategies affect the game, using a large amount of games for each strategy to try reduce the effect of anomalous results.
Thanks for the suggestion of Open Team Sheets, it is something I hadn't thought of and I agree that having it as an option would be a good addition to the project. I think best of 3 battles could be good, it would just increase the technical complexity of the AI, because I'd have to implement some way for it to decide whether or not to vary its strategy and leads. This could be done randomly, but ideally I'd rather it have actual strategy/reason behind it, I'm just not sure how that would be possible.
For the teams, I do plan on adding more, those three were honestly just what I came up with on the spot from what I have used and think is interesting/good, and were used for feedback, such as what you gave me. I should've been more clear about that, apologies. I do think that adding the most prevalent core restricted duos is necessary before the "completion" of the project, and I agree on the 10 team number you outlined, as it provides lots of variety for players to choose what team they like the most.
I'm not too worried about keeping it up to date to be honest. This is a 6-month or so school project that I doubt I'll do much with after this time period. I'd like to, especially if it turns out well, I just don't think I have the time.
Finally, thanks for the suggestion to ask for feedback from the technical projects forum; like I said I'm new and I don't know my way around the forums lol, so this is really helpful, thank you!
Yeah, you got it perfect, it's just a simple AI that is meant to mimic a human opponent.
Thanks as well for the recommendations, I'll make sure to have a look at those.