For your next video, would like to add that in vgc grass is actually a popular tera type due to its unique trait of resisting key types ground, electric, water and immunity to powder moves(spore, rage powder, etc) basically countering amoongus/not having to run safety goggles. Commonly seen on fire, water type mons or mons that want powder immunity: garchomp, indeenee, etc. It is sometimes seen in singles too for the same reason like glimora, iron moth, and even volcarona to get stab on giga drain as grass goes well offensively with fire hitting the water and rock that resist fire super effective
@@leaffinite2001 Which is why it's a good type. It resists extremely common types in Ground, Water, and Electric, while also being able to hit the extremely common types in Water, Ground, and Rock super effectively. And on the topic of Grass weaknesses, Bug and Poison are not very common at all offensively because they don't hit many common threats, Ice and Fire are pretty bad weaknesses to have because of how common they are as coverage moves on Water and Dragon Types, and Flying is a meh weakness considering Flying is a common typing, but there are usually better options for coverage attacks on your Pokémon, and thus Flying Type moves are only generally used on Pokémon who gain a stab bonus from it to make them stronger, in which case you don't keep your Grass type in against their Flying type, duh. Bug is worse because its offensive pressure is abysmal, being able to hit Psychic and Dark is bleh because those types have more common alternative weaknesses you could exploit, instead of using Bug, and those types aren't nearly as threating and Water, Rock, and Ground is. It also shares many of the same offensive resists, except worse because it's resisted by Fighting, Ghost, and Fairy in comparison to Grass being resisted by Grass, Bug, and Dragon. This is why lists like these don't really work, because they completely lack context on how common certain types are, and why those types are so commonly used in the first place.
I think grass as a whole type is bad on paper- but then you have amazing mons (rillaboom, amoonguss) and amazing moves learned by grass types (sleep powder, spore, rage powder) and the immunity to aforementioned moves as well as good STAB. So it's bad in a vacuum, but due to both metagame factors and general utility it actually really shines
I play lc and I love starting off games with early Tera grass glimlett to resist mons like diglett with 4x effective earth power and get either free rocks or an early kill
Based off type match ups this is good. But for the best team you’d need to factor in abilities, move pools, stats and what ever battle mechanics the game introduces for the current generation
True, would need to do that and factor terrastalizing, not to mention the 1v1 or 2v2 dynamic. Would definitely need to do more to “solve” team building. Also stuff like dark being immune to prankster and grass immune to powder moves, stuff like that.
I'm guessing you could feed back the results of the first run of those calcs as weights to recalc accounting for "what's popular". This would make water types better because they beat ground/ steel, for instance.
Just some corrections/add some nerdy detail 7:16 golem is actually not good in gen 1. It is currently NU. People used to think it was good as a zapdos check+explosion but as the meta evolved and gen 1 the body slam jank was discovered where it can’t para normal types, it dropped to uu but had no presence there and eventually nu where it is pretty decent. You were probably thinking of rhydon which is in ou and currently ranked 7th(the highest non normal/psychic type mon which are the two best types in gen 1 by a large margin) or gen 2 where golem is pretty good there with rapid spin, explosion stab earthquake to hit gengar trying to spin block and counters some snorlax sets(the undisputed best pokemon in ou and most agreed to be the best in ubers in gen 2). 9:45 skarmory while it was heavily used in stall, but also found in hyper offense a lot with its hazard lead set with stealth rock, spike, sturdy, custap berry. It guarantees 2 layers of hazards. Nowadays it isnt seen in stall anymore due to croviknight having a better stat spread and access to u-turn, pressure and pretty much only seen in hyper offense with its lead hazard set in ou 11:59 dragon was a terrible type in early gen. In gen 1,2 ou there was no dragons. Gen 3 only 2 dragons. It was only really broken in gen 4-5 due to the special/phy split and getting just better moves in general: dragon pulse, draco meteor, outrage(buffed from 90->120 bp). Also it wasnt used as a defensive type but more offensive since there was only one type, steel, that resisted it. Hence there were very few counters and you could just defeat through brute force(not to mention only 2 steel types had realiable/semi realiable recovery: skarmory-roost and ferrathorn-leech seed(lol grass)). Thats why drag-mag(dragons+magnezone) was a popular team archetype in gen 5(this strat wasnt relevant in gen 4 mainly because a lot of the good dragons were ban lol: garchomp, latios, salamence). The only dragon that was used defensively was latias, which only happen in gen 5 cause latios wasnt banned and parts of gen 4 when latios wasnt ban yet. 12:39 ice rock does exist, besides hisui avalugg. Aurorus. Which is terrible to no one’s surprise
I think this could go even deeper relating to the point about grass being the weakest vs Wolfeys opinion being bug, by assigning further values based on the individual typings that determine that value relating to their compound values. For example even though 2 individual types compound values might be equal, the values of the different typings they are super effective / resist might be lower or higher, breaking the tie in a way. It would make sense that resisting steel would be more valuable than resisting grass or bug, and it would be interesting to see that tie in to the data somehow
@@elzilcho222 you could make the variable whatever you want I’d assume as long as it holds a relationship with the other values, for example in this video we learned steel type is the best type based on the calculations of amounts of resistances/ weaknesses, not looking at what exactly we are resisting or weak to. Any type is worth 1 or worth -1. This was the start to assign initial values. Now, using that information we assign a new value to each type could just be 1-18 or more fractional than that might make more sense. But anyway we take our original data, and instead of every type being worth 1 point as a resistance or -1 point as a weakness, the amount of points added or subtracted are based on the ranking of the original more general calculation. Steel type showing up in resistance or weakness would be worth more or less points, and something like bug or grass showing up in resistance/weakness wouldn’t hold as much weight
@@elzilcho222 maybe grass could be 1 and the next type ranked could be worth 1.1 or -1.1, then 1.2 and -1.2 and so on. steel would be worth 2.7 points. Again these values would be based on the ranking in this video. I think this would factor in things such as a scenario where one typing that has 4 resistances, but happen to be the weakest 4 of all the types, ranks higher than a type with 3 resistances that happen to be the 3 strongest types. Having all types use a value of 1 would mean the former typing would be considered best when technically that may not be the case when factoring in what the types themselves that they resist/ are weak to are
Steel wasn't a good example. Steel is a rather rare attacking type besides to smack fairy and most things run Tera steel for the sole purpose of defense
It would probably be way too hard to do, but it would be really cool if you could have it weight the calculations based on the base stats of the pokemon that exist with those typings. I don’t know how much more efficient that would make it work, but it would be interesting
Excellent work! I had done similar work and arrived at similar conclusions, but hadn't extended my work to dual-types as extensively as you had (I mostly just looked at Fire dual-typings since Fire is my specialty, so seeing Fire/Ground rated so high was unsurprising to me). I do have a suggestion, however: since STAB is only a 1.5 multiplier, offensive typings should only be weighted half as much as defensive ones.
Good video. I have couple points to share about ranking types: 1. value of points could go deeper: since steel, water seem to be top tier, the type can hit it x1 or x2 damage gain a bit more point than 1 or 2. The opposite thing to worst type, the other types then gain less points against these bad types. 2. The total of offensive and defensive points could be calculated differently with database (computer things), such as average of Hp, atk, spd. For example, electric pokemons have the highest speed on average, so this type gain more offensive points.
i think this video is a great example of how different types and pokemon may seem good in a vacuum, but a pokemons strength is still ultimately determined by other factors, like stats, movepool, ability, synergy, role compression, and matchups with the current metagame. very interesting!
I think that the difference between yours and wolfie's is that not all relations ar created equal, for example since steel is so great having a positive match up against it is huge, ence the greater position of water in his list, and then since water becomes more important, grass hitting water makes it have a greater value etc... Edit also some types have special proprieties and specific moves that make It stronger
Me happy that this isn't a super popular video because I'm definitely about to try this. And it's already a well-known fact that steel type pokémon in fairy type pokémon are two of some of the best typings in the game and a steel fairy pokémon is probably one of the strongest pokémon you could probably ever run into which has a lot of them have been mythicals and in competitive teams so you're a tier list immediately looks accurate to me because you have them placed so highly in people figured this out doing the method similar to yours without using exact math They just paid attention to who has the least amount of weaknesses. So I feel like that without even needing to finish your video just looking at your tier list, It feels valid I love it. I didn't consider strategy I get to the end of your video and I'm like but there's no strategy in this It's just numbers to base typing but I wonder if I could create a strategy around this
You could probably apply what I did at the end of the video pretty well to gen 1 and 2 (and maybe the others to some extent), but gen 3 added abilities like sap sipper and now we have stuff like earth eater and adaptability that would throw the calc all out of wack if you’re looking for strategy. Maybe in the sequel I’ll include abilities and Tera types.
I was asking this same question about "what is the best team defensively/offensively?" but I also want to account for the moves that a given mon can learn to give their offensive coverage. Obviously, having 4 attacking moves of 4 different types isn't competitively viable in a lot of cases but it would just be cool to see the team that has the most defensive and offensive coverage.
5:00 - IIRC, part of the difference here is that his takes into account if a type is strong or weak against specific other strong or weak types, so it's kinda like he weighted certain types in his math, so his model is a little complex to model compared to the more straightforward one that you have.
Yeah grass type doesn't need a buff they don't just have grassy terrain they have the greatest access to healing out of all the types. Not only is it a great offensive type thanks to grassy terrain, they also have great longevity. It needs to be defensively weak as a type or else it wouldn't be balanced. It's also good against water and rock which are two of the best defensive types in the game. And it's also evident in their usage that they don't need a buff. If their usage were to fall off it might need a buff but it has enough positive qualities to be used enough. It's not bad. Bug type did need a buff when wolfey made that video. They did get a buff in gen 9 actually. Pounce can be learned by multiple pokemon it's physical 50 base power and always lower's the target's speed stat. Spidops has the special protect for this gen which is silk trap it lowers the opponent's speed when they make contact. Don't know how good that is but it makes scyther an attractive mon to use it outspeeds a lot of mons and enables allies to outspeed those mons as well even if they might be lower speed.
grass even with grassy terrain is weak offensively. On the other hand it's defensively really good due to natural resistance to common type like electric, ground or water. It's a mixed that got many synergy
Oh this I did this on a spreadsheet years ago! I love the graph interpretation I hadn’t considered that. The problem with this analysis is it assumes that all types are used in equal amounts. This isn’t true in practice. Theoretically you could actually weigh your graph results by giving better scores based on how good they are against the best types. Do this a few times and I’d be interested to see how the tiers change.
I remember doing something like this a while ago. (I also got Ice/Ground as rank 1 with a score of 311, but rock/ground is still 2nd with the same score of 306) but I used an array formula to sum the highest value between two types in a type chart with dual types on the defensive axis. Although one thing I recommend doing is accounting for no. of resists in the offensive end, since that has a bigger impact on making offensive combos really strong. Despite bolt/beam being rank 54 (250.5), it’s only resisted by 3 possible types which makes the coverage strong. While grass ground (260.5) has a higher score it’s far more resisted 34 (over 11x resists) so it’s not as potent. That’s more SE scores being weighted toward types that can hit 4x effectively. (I’ve yet to look at the sheet, so I’m using the values I found) One thing I find interesting is that ground/steel has the highest offensive and defensive score on my sheet as well.
I defende Wolfey's opinion about grass type. Grass have the powder and spore immunity that don't go in the calc, as well very important utility moves. They shiny more in a competitive way.
And in the case of not being ablr to use legendaries, you can use Kilowattrel instead of Thundurus, and Hisuian Goodra or Duraludon instead of Dialga. While they may be less useful, Kilowattrel is a good pokemon that can set Tailwind, Hisuian Goodra has really good defensive stats and Duraludon has a extremly good special attack. (and in my opinion, duraludon deserves more love)
A fun fact to throw in there, the only two Pokémon that collectively resist every type is hydreigon and aegislash (gholdengo too but I came up with this before gen nine)
Meowscarada is dark grass and its used A LOT in competitive. Dark means Prankster abilities do not effect it and grass protects it from spores it also hits hard
One reason I think are some differences between your tier list and wolfey's is that being super effective against a type is not as valuable as resisting it. Pokemon do more than just attack. The pokemon that DO just attack are often perfectly happy getting neutral damage against everything, so long as their attacks are not resisted. In competitive, super effective moves are most valuable on the coverage/nonSTAB moves as they can push past pokemon that resist your main attacking moves. So for example, Rock is a notoriously bad type. Rock types tend to struggle, with a few exceptions. Offensively, it is quite good, but defensively, it's quite bad. On paper this balances out, bu tin practice it is just very hard to switch on your rock types, and they can't take hits from many things. I think you would get results more in line with real life experience if being strong against a type was weighed less.high.
You ask if grass psychic exists, and exxegutor does exist! And its OU in gen 1! But... gen 1 has no steel, dark, or fairy, and also lots of types have no viable hitters like bug (4x dmg, but no viable attacks). So its weird Edit: not to mention some types resistances dont even work correctly.
You could probably come up with a formula to scale the value of these multipliers. Having a 2x resist to ground, for example, is more valuable than a 2x resist to grass, and an immunity to dragon is more valuable than an immunity to poison. Immunities are also better than resistances, which I don’t think your model took into account. Something like “the value of the node is scaled based on the ranking of the defending type”. Might get you closer to the real world versions.
One thing I would love to see is it going a layer deeper to account for some context. For example being able to hit steel types for super effective damage is more valuable than hitting bugs for super effective damage. Simular resisting ground and rock is more valuable than resisting normal bug
So I would love to see this system turned into something you could put any type into. Like you put in fire ghost and it will give you the mathematically best team typing wise to support that Pokémon.
A greedy algorithm wouldn't work for a perfect team, so picking the best new one each time doesn't give the mathematically optimal variant. Also you already need a whole team picked to check the attack/defense values by your model. After I ran a program that tries all the combinations (roughly 10^14 simple operations, 25 minutes on my PC), it gave me a team like this: Normal/Flying, Fighting/Ground, Fighting/Dark, Rock/Fairy, Ghost/Grass and Steel/Fire.
I used a similar spreadsheet and got slightly different results. I got water ground as the best type combination. I computed the eigenvector of a matrix of winning type matchups rather than compute the difference between offensive and defensive type matchups.
The main difference between your ranking and mine is Ghost Psychic and Fighting Rock. My ranking considers Ghost Psychic as a middle tier type combination whereas your ranking considers the combination as really bad. My ranking considers Fighting Rock as a bad combination whereas your ranking considers it a middle tier type combination.
I have a document I've been working with that I want to show you. It rebalances the types and considers things like movepools, abilities, average stats, matchups vs common types (way more water types than ice types for ex.), the weights of weaknesses, etc (as well as thematics of course). I would love for you to look at it if you ever get the chance. Thanks.
I believe grass is regarded as higher tier because there are few mono grass mons and those do pretty bad comparatively. People associate with other types subconsciously.
Grass is immune to powder moves though. In the same way, Dark is immune to moves effected by the ability Prankster. This is a type thing, and is accounted for by Wolfey. I like the video, but there is more to types then just how they effect each other.
I've noticed a glaring oversight in your data here: After several hours of recreating your code and debugging inconsistencies, I discovered that an error I was experiencing was actually not a bug in my code, but in yours. Your code errantly has Ice marked as dealing 1x damage to Dragon, rather than 2x. This skews pretty much all of the data shown in the video.
You know, grass is a bad type. And people tend to say it’s because it has air of weaknesses. But I think the reason that it ranks low mathetically is actually because it has a high amount of NVE .
One thing this doesn't account for is coverage moves. Your spreadsheet works perfectly if we assume pokémon are always attacking with one of their STAB moves, which we know is not true. You are incentivized, through STAB, to run moves from the same type as the pokémon using them, but you don't HAVE TO, and even if you do, you can always suplement a lacking offensive typing with a varied movepool. While, when it comes to defensive relationships, you're pretty much locked to your own type's properties. With this in mind, I would argue that a strong defensive typing is a bit more valuable than a strong offensive typing, since defense is more intrinsic/deterministic. That's probably why Ice feels so bad. Every Water type can use Ice Beam and benefit from Ice's offensive pros, but how do you run around Ice's defensive cons? You don't.
very late but as several other ppl seemed to suggest here, there actually isnt a difference between the theoretical and practical best types this network is just missing a key aspect which is valuing different resistances/weakness or effectiveness more or less depending on how good the other type is. grass is an excellent type because it's good into ground and water being the best example of that, grass exists to be a counter to statistically good types. the only practical way i see to do that without creating infinite feedback loops is to use this network as a base, then evaluate each type again where you multiply the line value by the previous value of the other type, and im not sure at all but i think if you did that infinitely it would converge on the correct rankings
This is a cool math excersize but I found that everyone always has a flawed approach when it comes to optimal typings. People always try to blindly add total weaknesses and resistances without looking at how they line up. Water resisting steel isn’t nearly as useful as fire because fire can hit it super effectively in return for example. How I teambuild optimally is only counting a type as being another if it both resists and is super effective against it. When you do this the optimal team comp is grass dark ground fire fairy rock and poison/steel (you have an option to resists every type and simultaneously hit super effectively except normal because it’s not possible)
Idk man, you end up restricting your teambuilding way too much in this way. As long as your anwer can beat what it is supposed to, who cares if it is causing super-effective-STAB damage or not. Thats what coverage is there for.
@@walaceandrade4046 because a 3x damage multiplier is much bigger than 2x. And yeah this is just what I think is optimal I don’t limit my teambuilding to just these eight types
I hate how unthematic ground/rock being the number 1 attacking type and the 3rd worst defending type. It should be the other way around. We’ll, actually rock steel should be the best defending type. But rock ground should but super defensive too
Scratches me head and ape grunts. Translation:"I wasn't paying attention to the maths" I thought you'd like do a breakdown of the typing chart. Nice presentation though 🤣😅
I wonder if you could figure an adjustable combined values reflective of speed, ie the greater speed the more important offense is and the less important defense is and vice versa O.o
I remember opening your doc and using it a few times, but it seems I never made a copy! Does anyone have a copy I can...copy? Then I don't need to rely on anyone else to be able to view it after I make a copy!
That team on the thumbnail doesn’t make sense as the best because there’s 2 type overlaps (steel with duraludon and excadrill, ghost with hisuian zoroark and sableye) oh and fairy steel is a better dual type than ground steel. If a Pokémon world champion (WofleyVGC) says so, then it’s facts.
I've actually recreated most of it myself now, see my other comment Edit: omg please help me I CANNOT calculate the optimal team to save my life I have been at this for HOURS
This is interesting but it's far too simplistic to actually result in a great team. Base stats and abilities are just too weighty in terms of actual play. I've done something similar to this before but went a couple of steps further and factored in statistics on move types common in the metagame - i.e. if everything is running ground, water and fighting coverage moves, your ground/steel type is not going to have a great time..
For your next video, would like to add that in vgc grass is actually a popular tera type due to its unique trait of resisting key types ground, electric, water and immunity to powder moves(spore, rage powder, etc) basically countering amoongus/not having to run safety goggles. Commonly seen on fire, water type mons or mons that want powder immunity: garchomp, indeenee, etc. It is sometimes seen in singles too for the same reason like glimora, iron moth, and even volcarona to get stab on giga drain as grass goes well offensively with fire hitting the water and rock that resist fire super effective
Its shit conceptually except its usable against the more generically good stuff.
@@leaffinite2001 Which is why it's a good type. It resists extremely common types in Ground, Water, and Electric, while also being able to hit the extremely common types in Water, Ground, and Rock super effectively.
And on the topic of Grass weaknesses, Bug and Poison are not very common at all offensively because they don't hit many common threats, Ice and Fire are pretty bad weaknesses to have because of how common they are as coverage moves on Water and Dragon Types, and Flying is a meh weakness considering Flying is a common typing, but there are usually better options for coverage attacks on your Pokémon, and thus Flying Type moves are only generally used on Pokémon who gain a stab bonus from it to make them stronger, in which case you don't keep your Grass type in against their Flying type, duh.
Bug is worse because its offensive pressure is abysmal, being able to hit Psychic and Dark is bleh because those types have more common alternative weaknesses you could exploit, instead of using Bug, and those types aren't nearly as threating and Water, Rock, and Ground is. It also shares many of the same offensive resists, except worse because it's resisted by Fighting, Ghost, and Fairy in comparison to Grass being resisted by Grass, Bug, and Dragon.
This is why lists like these don't really work, because they completely lack context on how common certain types are, and why those types are so commonly used in the first place.
@@michaelbowman6684 right
I think grass as a whole type is bad on paper- but then you have amazing mons (rillaboom, amoonguss) and amazing moves learned by grass types (sleep powder, spore, rage powder) and the immunity to aforementioned moves as well as good STAB. So it's bad in a vacuum, but due to both metagame factors and general utility it actually really shines
I play lc and I love starting off games with early Tera grass glimlett to resist mons like diglett with 4x effective earth power and get either free rocks or an early kill
Based off type match ups this is good. But for the best team you’d need to factor in abilities, move pools, stats and what ever battle mechanics the game introduces for the current generation
And the META
True, would need to do that and factor terrastalizing, not to mention the 1v1 or 2v2 dynamic. Would definitely need to do more to “solve” team building. Also stuff like dark being immune to prankster and grass immune to powder moves, stuff like that.
I mean yeah but thats why you cant ever really solve teambuilding. Thats the only reason pokemon is fun
I'm guessing you could feed back the results of the first run of those calcs as weights to recalc accounting for "what's popular". This would make water types better because they beat ground/ steel, for instance.
Just some corrections/add some nerdy detail
7:16 golem is actually not good in gen 1. It is currently NU. People used to think it was good as a zapdos check+explosion but as the meta evolved and gen 1 the body slam jank was discovered where it can’t para normal types, it dropped to uu but had no presence there and eventually nu where it is pretty decent. You were probably thinking of rhydon which is in ou and currently ranked 7th(the highest non normal/psychic type mon which are the two best types in gen 1 by a large margin) or gen 2 where golem is pretty good there with rapid spin, explosion stab earthquake to hit gengar trying to spin block and counters some snorlax sets(the undisputed best pokemon in ou and most agreed to be the best in ubers in gen 2).
9:45 skarmory while it was heavily used in stall, but also found in hyper offense a lot with its hazard lead set with stealth rock, spike, sturdy, custap berry. It guarantees 2 layers of hazards. Nowadays it isnt seen in stall anymore due to croviknight having a better stat spread and access to u-turn, pressure and pretty much only seen in hyper offense with its lead hazard set in ou
11:59 dragon was a terrible type in early gen. In gen 1,2 ou there was no dragons. Gen 3 only 2 dragons. It was only really broken in gen 4-5 due to the special/phy split and getting just better moves in general: dragon pulse, draco meteor, outrage(buffed from 90->120 bp). Also it wasnt used as a defensive type but more offensive since there was only one type, steel, that resisted it. Hence there were very few counters and you could just defeat through brute force(not to mention only 2 steel types had realiable/semi realiable recovery: skarmory-roost and ferrathorn-leech seed(lol grass)). Thats why drag-mag(dragons+magnezone) was a popular team archetype in gen 5(this strat wasnt relevant in gen 4 mainly because a lot of the good dragons were ban lol: garchomp, latios, salamence). The only dragon that was used defensively was latias, which only happen in gen 5 cause latios wasnt banned and parts of gen 4 when latios wasnt ban yet.
12:39 ice rock does exist, besides hisui avalugg. Aurorus. Which is terrible to no one’s surprise
This can be very good for building a team that beats the meta. Love it. It’s great for team-building in general. Spreadsheets are awesome.
Thank you!
I think this could go even deeper relating to the point about grass being the weakest vs Wolfeys opinion being bug, by assigning further values based on the individual typings that determine that value relating to their compound values. For example even though 2 individual types compound values might be equal, the values of the different typings they are super effective / resist might be lower or higher, breaking the tie in a way. It would make sense that resisting steel would be more valuable than resisting grass or bug, and it would be interesting to see that tie in to the data somehow
How would you determine what those further values are? Like, just from the compound value ranking? How would you turn the ranking into a number?
@@elzilcho222 you could make the variable whatever you want I’d assume as long as it holds a relationship with the other values, for example in this video we learned steel type is the best type based on the calculations of amounts of resistances/ weaknesses, not looking at what exactly we are resisting or weak to. Any type is worth 1 or worth -1. This was the start to assign initial values. Now, using that information we assign a new value to each type could just be 1-18 or more fractional than that might make more sense. But anyway we take our original data, and instead of every type being worth 1 point as a resistance or -1 point as a weakness, the amount of points added or subtracted are based on the ranking of the original more general calculation. Steel type showing up in resistance or weakness would be worth more or less points, and something like bug or grass showing up in resistance/weakness wouldn’t hold as much weight
@@elzilcho222 maybe grass could be 1 and the next type ranked could be worth 1.1 or -1.1, then 1.2 and -1.2 and so on. steel would be worth 2.7 points. Again these values would be based on the ranking in this video. I think this would factor in things such as a scenario where one typing that has 4 resistances, but happen to be the weakest 4 of all the types, ranks higher than a type with 3 resistances that happen to be the 3 strongest types. Having all types use a value of 1 would mean the former typing would be considered best when technically that may not be the case when factoring in what the types themselves that they resist/ are weak to are
Steel wasn't a good example. Steel is a rather rare attacking type besides to smack fairy and most things run Tera steel for the sole purpose of defense
Content like this deserves so much more than 80 subs, good luck to you man.
Thank you brother, I’ll keep the grind up!
It would probably be way too hard to do, but it would be really cool if you could have it weight the calculations based on the base stats of the pokemon that exist with those typings. I don’t know how much more efficient that would make it work, but it would be interesting
Excellent work! I had done similar work and arrived at similar conclusions, but hadn't extended my work to dual-types as extensively as you had (I mostly just looked at Fire dual-typings since Fire is my specialty, so seeing Fire/Ground rated so high was unsurprising to me). I do have a suggestion, however: since STAB is only a 1.5 multiplier, offensive typings should only be weighted half as much as defensive ones.
Good video.
I have couple points to share about ranking types:
1. value of points could go deeper: since steel, water seem to be top tier, the type can hit it x1 or x2 damage gain a bit more point than 1 or 2. The opposite thing to worst type, the other types then gain less points against these bad types.
2. The total of offensive and defensive points could be calculated differently with database (computer things), such as average of Hp, atk, spd. For example, electric pokemons have the highest speed on average, so this type gain more offensive points.
i think this video is a great example of how different types and pokemon may seem good in a vacuum, but a pokemons strength is still ultimately determined by other factors, like stats, movepool, ability, synergy, role compression, and matchups with the current metagame. very interesting!
I think that the difference between yours and wolfie's is that not all relations ar created equal, for example since steel is so great having a positive match up against it is huge, ence the greater position of water in his list, and then since water becomes more important, grass hitting water makes it have a greater value etc...
Edit also some types have special proprieties and specific moves that make It stronger
Me happy that this isn't a super popular video because I'm definitely about to try this. And it's already a well-known fact that steel type pokémon in fairy type pokémon are two of some of the best typings in the game and a steel fairy pokémon is probably one of the strongest pokémon you could probably ever run into which has a lot of them have been mythicals and in competitive teams so you're a tier list immediately looks accurate to me because you have them placed so highly in people figured this out doing the method similar to yours without using exact math They just paid attention to who has the least amount of weaknesses. So I feel like that without even needing to finish your video just looking at your tier list, It feels valid I love it. I didn't consider strategy I get to the end of your video and I'm like but there's no strategy in this It's just numbers to base typing but I wonder if I could create a strategy around this
You could probably apply what I did at the end of the video pretty well to gen 1 and 2 (and maybe the others to some extent), but gen 3 added abilities like sap sipper and now we have stuff like earth eater and adaptability that would throw the calc all out of wack if you’re looking for strategy. Maybe in the sequel I’ll include abilities and Tera types.
You're not even a nerd you're just a smart person bro HUGE RESPECT
I was asking this same question about "what is the best team defensively/offensively?" but I also want to account for the moves that a given mon can learn to give their offensive coverage. Obviously, having 4 attacking moves of 4 different types isn't competitively viable in a lot of cases but it would just be cool to see the team that has the most defensive and offensive coverage.
5:00 - IIRC, part of the difference here is that his takes into account if a type is strong or weak against specific other strong or weak types, so it's kinda like he weighted certain types in his math, so his model is a little complex to model compared to the more straightforward one that you have.
Yes, that is a VERY cute baby Skarmory😅
The teeny meow was adorable
Yeah grass type doesn't need a buff they don't just have grassy terrain they have the greatest access to healing out of all the types. Not only is it a great offensive type thanks to grassy terrain, they also have great longevity. It needs to be defensively weak as a type or else it wouldn't be balanced. It's also good against water and rock which are two of the best defensive types in the game. And it's also evident in their usage that they don't need a buff. If their usage were to fall off it might need a buff but it has enough positive qualities to be used enough. It's not bad.
Bug type did need a buff when wolfey made that video. They did get a buff in gen 9 actually. Pounce can be learned by multiple pokemon it's physical 50 base power and always lower's the target's speed stat. Spidops has the special protect for this gen which is silk trap it lowers the opponent's speed when they make contact. Don't know how good that is but it makes scyther an attractive mon to use it outspeeds a lot of mons and enables allies to outspeed those mons as well even if they might be lower speed.
Grass is also immune to powder moves! Thanks for watching
grass even with grassy terrain is weak offensively. On the other hand it's defensively really good due to natural resistance to common type like electric, ground or water. It's a mixed that got many synergy
Hold up, what? Rock is absolutely not one of the best defensive types in the game.
It's excellent offensively, but defensively Rock is horrible.
Oh this I did this on a spreadsheet years ago! I love the graph interpretation I hadn’t considered that.
The problem with this analysis is it assumes that all types are used in equal amounts. This isn’t true in practice. Theoretically you could actually weigh your graph results by giving better scores based on how good they are against the best types. Do this a few times and I’d be interested to see how the tiers change.
I have no real comment except great video, I can tell it took a ton of effort and you presented it in a very digestible manner
I remember doing something like this a while ago.
(I also got Ice/Ground as rank 1 with a score of 311, but rock/ground is still 2nd with the same score of 306) but I used an array formula to sum the highest value between two types in a type chart with dual types on the defensive axis.
Although one thing I recommend doing is accounting for no. of resists in the offensive end, since that has a bigger impact on making offensive combos really strong. Despite bolt/beam being rank 54 (250.5), it’s only resisted by 3 possible types which makes the coverage strong. While grass ground (260.5) has a higher score it’s far more resisted 34 (over 11x resists) so it’s not as potent.
That’s more SE scores being weighted toward types that can hit 4x effectively. (I’ve yet to look at the sheet, so I’m using the values I found)
One thing I find interesting is that ground/steel has the highest offensive and defensive score on my sheet as well.
I defende Wolfey's opinion about grass type. Grass have the powder and spore immunity that don't go in the calc, as well very important utility moves. They shiny more in a competitive way.
And in the case of not being ablr to use legendaries, you can use Kilowattrel instead of Thundurus, and Hisuian Goodra or Duraludon instead of Dialga. While they may be less useful, Kilowattrel is a good pokemon that can set Tailwind, Hisuian Goodra has really good defensive stats and Duraludon has a extremly good special attack. (and in my opinion, duraludon deserves more love)
A fun fact to throw in there, the only two Pokémon that collectively resist every type is hydreigon and aegislash (gholdengo too but I came up with this before gen nine)
Meowscarada is dark grass and its used A LOT in competitive. Dark means Prankster abilities do not effect it and grass protects it from spores it also hits hard
One reason I think are some differences between your tier list and wolfey's is that being super effective against a type is not as valuable as resisting it. Pokemon do more than just attack. The pokemon that DO just attack are often perfectly happy getting neutral damage against everything, so long as their attacks are not resisted. In competitive, super effective moves are most valuable on the coverage/nonSTAB moves as they can push past pokemon that resist your main attacking moves.
So for example, Rock is a notoriously bad type. Rock types tend to struggle, with a few exceptions. Offensively, it is quite good, but defensively, it's quite bad. On paper this balances out, bu tin practice it is just very hard to switch on your rock types, and they can't take hits from many things.
I think you would get results more in line with real life experience if being strong against a type was weighed less.high.
You ask if grass psychic exists, and exxegutor does exist! And its OU in gen 1! But... gen 1 has no steel, dark, or fairy, and also lots of types have no viable hitters like bug (4x dmg, but no viable attacks). So its weird
Edit: not to mention some types resistances dont even work correctly.
You could probably come up with a formula to scale the value of these multipliers. Having a 2x resist to ground, for example, is more valuable than a 2x resist to grass, and an immunity to dragon is more valuable than an immunity to poison. Immunities are also better than resistances, which I don’t think your model took into account. Something like “the value of the node is scaled based on the ranking of the defending type”. Might get you closer to the real world versions.
One thing I would love to see is it going a layer deeper to account for some context. For example being able to hit steel types for super effective damage is more valuable than hitting bugs for super effective damage. Simular resisting ground and rock is more valuable than resisting normal bug
So I would love to see this system turned into something you could put any type into. Like you put in fire ghost and it will give you the mathematically best team typing wise to support that Pokémon.
Some people might try to use a bug/grass in parasect for spore, or, not for competitive, leavanny learns false swipe for shiny hunting
Love the Chao garden music
A greedy algorithm wouldn't work for a perfect team, so picking the best new one each time doesn't give the mathematically optimal variant. Also you already need a whole team picked to check the attack/defense values by your model.
After I ran a program that tries all the combinations (roughly 10^14 simple operations, 25 minutes on my PC), it gave me a team like this: Normal/Flying, Fighting/Ground, Fighting/Dark, Rock/Fairy, Ghost/Grass and Steel/Fire.
I did this with dual vs single types, it matched up way more like wolfeys list
I used a similar spreadsheet and got slightly different results. I got water ground as the best type combination.
I computed the eigenvector of a matrix of winning type matchups rather than compute the difference between offensive and defensive type matchups.
The main difference between your ranking and mine is Ghost Psychic and Fighting Rock. My ranking considers Ghost Psychic as a middle tier type combination whereas your ranking considers the combination as really bad. My ranking considers Fighting Rock as a bad combination whereas your ranking considers it a middle tier type combination.
Great video! Makes me want to pay more attention to math class
2 seconds in and an automatic like from me for hearing chao daycare music. Also this is very well put together video
I have a document I've been working with that I want to show you. It rebalances the types and considers things like movepools, abilities, average stats, matchups vs common types (way more water types than ice types for ex.), the weights of weaknesses, etc (as well as thematics of course). I would love for you to look at it if you ever get the chance. Thanks.
I applied the work you put in
Thank you, let me know how well (or not we’ll) it works out!
I like your deep analysis, very useful
defensively is more important as the aim of the game is to not be KOed. And Monsters are not limited to only using attacks that are their type
Azu gets sap sipper and thundurus gets volt absorb so that final team has even more inmunities xD
Grass is immune to spore + rage powder.... and has access to spore and rage powder ;) interesting to see that such a thing is so impactful
Can't Spore Grass.
Can't Taunt Dark.
Can't Burn Fire, ext...
Factor this in
I believe grass is regarded as higher tier because there are few mono grass mons and those do pretty bad comparatively. People associate with other types subconsciously.
3:53 omg kitty!
bottom left corner btw
This is sick
Thanks!
The types don’t do everything. You can have an excellent type, but a lot of counters in the metagame.
Grass is immune to powder moves though. In the same way, Dark is immune to moves effected by the ability Prankster. This is a type thing, and is accounted for by Wolfey. I like the video, but there is more to types then just how they effect each other.
12:40 i think aurorous is also ice rock.
I've noticed a glaring oversight in your data here:
After several hours of recreating your code and debugging inconsistencies, I discovered that an error I was experiencing was actually not a bug in my code, but in yours. Your code errantly has Ice marked as dealing 1x damage to Dragon, rather than 2x. This skews pretty much all of the data shown in the video.
Alright, people, yes, but you've got to subscribe to the channel when the algorithm does this
I can't imagine how long this must have taken!
great stuff man
You know, grass is a bad type. And people tend to say it’s because it has air of weaknesses. But I think the reason that it ranks low mathetically is actually because it has a high amount of NVE .
Celebi is Psychic/Grass i think
And Grass/Ice is in Diamond and Pearl
One thing this doesn't account for is coverage moves. Your spreadsheet works perfectly if we assume pokémon are always attacking with one of their STAB moves, which we know is not true. You are incentivized, through STAB, to run moves from the same type as the pokémon using them, but you don't HAVE TO, and even if you do, you can always suplement a lacking offensive typing with a varied movepool. While, when it comes to defensive relationships, you're pretty much locked to your own type's properties. With this in mind, I would argue that a strong defensive typing is a bit more valuable than a strong offensive typing, since defense is more intrinsic/deterministic.
That's probably why Ice feels so bad. Every Water type can use Ice Beam and benefit from Ice's offensive pros, but how do you run around Ice's defensive cons? You don't.
very late but as several other ppl seemed to suggest here, there actually isnt a difference between the theoretical and practical best types this network is just missing a key aspect which is valuing different resistances/weakness or effectiveness more or less depending on how good the other type is. grass is an excellent type because it's good into ground and water being the best example of that, grass exists to be a counter to statistically good types. the only practical way i see to do that without creating infinite feedback loops is to use this network as a base, then evaluate each type again where you multiply the line value by the previous value of the other type, and im not sure at all but i think if you did that infinitely it would converge on the correct rankings
This is a cool math excersize but I found that everyone always has a flawed approach when it comes to optimal typings. People always try to blindly add total weaknesses and resistances without looking at how they line up. Water resisting steel isn’t nearly as useful as fire because fire can hit it super effectively in return for example. How I teambuild optimally is only counting a type as being another if it both resists and is super effective against it. When you do this the optimal team comp is grass dark ground fire fairy rock and poison/steel (you have an option to resists every type and simultaneously hit super effectively except normal because it’s not possible)
Idk man, you end up restricting your teambuilding way too much in this way. As long as your anwer can beat what it is supposed to, who cares if it is causing super-effective-STAB damage or not. Thats what coverage is there for.
@@walaceandrade4046 because a 3x damage multiplier is much bigger than 2x. And yeah this is just what I think is optimal I don’t limit my teambuilding to just these eight types
I never sub just because a youtuber tells me to but this was really interesting, so might aswell
I hate how unthematic ground/rock being the number 1 attacking type and the 3rd worst defending type. It should be the other way around. We’ll, actually rock steel should be the best defending type. But rock ground should but super defensive too
Scratches me head and ape grunts.
Translation:"I wasn't paying attention to the maths"
I thought you'd like do a breakdown of the typing chart.
Nice presentation though 🤣😅
Keep up the good work
Thanks, I’ll try!
15:40 *calyrex and abomasnow show up
18:25 I thought you had grass/fire as the last type combo, why'd you change it to dragon/steel?
I wonder if you could figure an adjustable combined values reflective of speed, ie the greater speed the more important offense is and the less important defense is and vice versa O.o
140th subscriber!
I remember opening your doc and using it a few times, but it seems I never made a copy! Does anyone have a copy I can...copy? Then I don't need to rely on anyone else to be able to view it after I make a copy!
what about instead of weighting it against the current best type, weighting it against every type at once
Not sure what you mean, each type is weighed against every type, that’s how we get the values next to every type on the spreadsheet.
Are you referring to the eigenvector of the graph?
That team on the thumbnail doesn’t make sense as the best because there’s 2 type overlaps (steel with duraludon and excadrill, ghost with hisuian zoroark and sableye) oh and fairy steel is a better dual type than ground steel. If a Pokémon world champion (WofleyVGC) says so, then it’s facts.
awesome!!
Thank you!
Good job bebe 🎉
Thanks bubu!
What about.. held items?
Who is the best statistical pokemon?
Your mom
Mega Rayquaza.
Is there any chance you'll release the code for this version? It would be interesting to calculate the optimal teammates for each type combination
I've actually recreated most of it myself now, see my other comment
Edit: omg please help me I CANNOT calculate the optimal team to save my life I have been at this for HOURS
It has been over a month please help me
This is interesting but it's far too simplistic to actually result in a great team. Base stats and abilities are just too weighty in terms of actual play. I've done something similar to this before but went a couple of steps further and factored in statistics on move types common in the metagame - i.e. if everything is running ground, water and fighting coverage moves, your ground/steel type is not going to have a great time..
people are sleeping on defensive Tera Electric
When everything runs earthquake, you gotta have some big balls to run defensive tera eletric. XD
Unfortunately actually using that team would pretty quickly make you realize that there's way more to Pokemon than type
But steel is weak to fire… and ground…
You didn’t account for STAB
❤
I went to like this video and I was the 1000th person to do so
Dragon resist 4 types not 3
Woops, forgot about electric, you’re right!
@@elzilcho222 are you replying on an alternate account on purpose? love the video btw
@@SirTrapz Gosu garage my me and my friends, this is my personal account. Thank you! comments like this make me want to keep making videos.
12:39 *Hisuian not galarian 🤓
What if i dont have thundurus or hisuian zourark?
Or exadrill or duraludon?
Or sableye?
Wanted to add enjoyed the video and subbed 👍
Would love to see the results with some battles