Typing Rating

Status
Not open for further replies.
#27
X-Act, if you're happy with the ratings, it's time for part 2:

I'd like to see if there is some way to incorporate this, or some derivative of this, into the CAP process. I really have no clue how to do that. But here's my motivation -- our current type selection process is a total crapshoot. We just randomly throw out potential types, with the hope that it will support our concept, and we take it from there. If we could somehow put a more analytical framework underneath our type discussions and selection process -- perhaps we could add a bit more structure and reason to our project.

I have a lot of random thoughts about this right now, and I don't want to go in a million directions at once. But, perhaps others can think about this stuff too and we might be able to come up with something together. Or maybe not. At the very least, it could be an interesting discussion!

Once again -- great work X-Act! I love when you present this kind of thing. It's always fascinating to see numbers that support our common "gut feelings" about this game we play.
Answer DougJustDoug's question. Anyone thinght of anything?
 
#28
I don't know if it's possible but if we could get moves and abilities into some standard form to evaluate, we could derive a total statistic to rank every Pokemon which could be very powerful for new CAP creation.
 
#29
HI X-Act,
It is a very good work from you. The ratings are absolutely a very interesting type of data that would help select a better Pokemon. It would make the work of selecting a Pokemon a much easier process. It was very kind of you to share such useful information with the others. Thank you for the post and keep the good work going. Stay connected.
 

X-Act

np: Biffy Clyro - Shock Shock
is a Site Staff Alumnusis a Programmer Alumnusis a Live Chat Contributor Alumnusis a Researcher Alumnusis a CAP Contributor Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnusis an Administrator Alumnus
#30
I don't know if it's possible but if we could get moves and abilities into some standard form to evaluate, we could derive a total statistic to rank every Pokemon which could be very powerful for new CAP creation.
That's what I wanted to do for the past two years or so. It would not only be good for CAP; it would also be good to aid the suspect process.

I can easily assign a rating for each move that does damage without much problems. It's the moves that don't do damage that are a problem.

Also, there's Ability that needs to be taken care of too. And, as I said, there's the problem of evaluating numerically moves like Confuse Ray, Trick, Stealth Rock, Toxic Spikes, Toxic, Sleep Powder, ... basically all moves that don't do damage.

Not to mention rating numerically abilities like Keen Eye, Own Tempo, Trace, Rock Head, etc.

I don't think that such a rating is impossible, but it is surely very difficult.
 
#31
I think the adjusted ratings may be a bit off. I'm getting 72.0188 instead of 76.8699 for water/none, 58.1208 instead of 63.0470 for dark/ghost, and 72.5683 instead of 78.0682 for normal/none, for instance.

Also, have you thought of a way to integrate stealth rock resistances, weather immunities and the like into the ratings?
 

X-Act

np: Biffy Clyro - Shock Shock
is a Site Staff Alumnusis a Programmer Alumnusis a Live Chat Contributor Alumnusis a Researcher Alumnusis a CAP Contributor Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnusis an Administrator Alumnus
#32
I think the adjusted ratings may be a bit off. I'm getting 72.0188 instead of 76.8699 for water/none, 58.1208 instead of 63.0470 for dark/ghost, and 72.5683 instead of 78.0682 for normal/none, for instance.

Also, have you thought of a way to integrate stealth rock resistances, weather immunities and the like into the ratings?
It depends on the method you're using to calculate them. How are you calculating them?

These numbers don't take any move information into account. For instance, Fighting should technically score a higher OTR because Fighting moves tend to have a higher Base Power than, say, Ghost moves. So no, I haven't incorporated any move information in the ratings shown.
 

zarator

^_^
is a Forum Moderatoris a Community Contributoris a Contributor Alumnus
Moderator
#33
It depends on the method you're using to calculate them. How are you calculating them?

These numbers don't take any move information into account. For instance, Fighting should technically score a higher OTR because Fighting moves tend to have a higher Base Power than, say, Ghost moves. So no, I haven't incorporated any move information in the ratings shown.
I don't think you should incorporate such an information, especially if it is thought for CAP use. Not only, as you said, being Fighting (for example) does not mean that you get Close Combat, CAP as a long history of move creation which would alter drastically the OTR should you take moves into account (think of how much weight would have Paleo Wave compared to Ancientpower and Power Gem, for example). I'm all for armogohma said about SR resistance, weather immunity and so on, though, especially the former. A 4X Rock Weakness must weigh a lot more on the DTR than , say, a 4X Ice move, despite Rock and Ice being similar types OTR-wise.
 

X-Act

np: Biffy Clyro - Shock Shock
is a Site Staff Alumnusis a Programmer Alumnusis a Live Chat Contributor Alumnusis a Researcher Alumnusis a CAP Contributor Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnusis an Administrator Alumnus
#34
I don't think you should incorporate such an information, especially if it is thought for CAP use. Not only, as you said, being Fighting (for example) does not mean that you get Close Combat, CAP as a long history of move creation which would alter drastically the OTR should you take moves into account (think of how much weight would have Paleo Wave compared to Ancientpower and Power Gem, for example). I'm all for armogohma said about SR resistance, weather immunity and so on, though, especially the former. A 4X Rock Weakness must weigh a lot more on the DTR than , say, a 4X Ice move, despite Rock and Ice being similar types OTR-wise.
I would gladly do this... however I would like a nice list of commonly-used attacking moves, their accuracy, their base power and any side-effect they have, sorted by their types. I think Deck Knight has something already about this, and maybe you have too, since you do the Move type frequency stats in Stark.

About Rock weakness etc., I've already weighted that accordingly in the DTR. If you think about it, there are about as many Ice-weak Pokemon than Rock-weak ones (if not more!), so, if an Ice-type Stealth Rock (Stealth Ice?) move was to be invented, it would pretty much be as effective to the metagame as Stealth Rock is.
 

zarator

^_^
is a Forum Moderatoris a Community Contributoris a Contributor Alumnus
Moderator
#35
I would gladly do this... however I would like a nice list of commonly-used attacking moves, their accuracy, their base power and any side-effect they have, sorted by their types. I think Deck Knight has something already about this, and maybe you have too, since you do the Move type frequency stats in Stark.

About Rock weakness etc., I've already weighted that accordingly in the DTR. If you think about it, there are about as many Ice-weak Pokemon than Rock-weak ones (if not more!), so, if an Ice-type Stealth Rock (Stealth Ice?) move was to be invented, it would pretty much be as effective to the metagame as Stealth Rock is.
http://s1.zetaboards.com/PokemonDraftProject/topic/1018451/1/

The site itself is pretty void and abandoned, but I think you will find this post of your interest. Some choices of moves are actually debatable, but it is nonetheless a good base.
 
#36
A Stealth Ice would ruin the metagame, Dragon-wise... Might as well make a Fire one and get done with Scizor and Jirachi
 
#37
X-Act, I do have a couple of questions that I wish to ask in regards to the typing rating presented here.

First of all, in your offensive type rating, you mentioned that when you use the formula, you should get a number somewhere in between 1.0844 and 1.1304. I tried the bug/flying typing just to see what I would end up with and got approximately 1.1226. It is closer to the upper end of that range, that much I can tell. The question that arises for me is this: is being closer to the maximum give you a good or a bad rating?

Secondly, while you do not appear to currently have an overall type rating formula (as in a formula that takes the offensive and defensive type ratings and outputs some number), would such a formula be practical for determining how good a Pokemon's type is overall?
 

X-Act

np: Biffy Clyro - Shock Shock
is a Site Staff Alumnusis a Programmer Alumnusis a Live Chat Contributor Alumnusis a Researcher Alumnusis a CAP Contributor Alumnusis a Tiering Contributor Alumnusis a Contributor Alumnusis an Administrator Alumnus
#38
X-Act, I do have a couple of questions that I wish to ask in regards to the typing rating presented here.

First of all, in your offensive type rating, you mentioned that when you use the formula, you should get a number somewhere in between 1.0844 and 1.1304. I tried the bug/flying typing just to see what I would end up with and got approximately 1.1226. It is closer to the upper end of that range, that much I can tell. The question that arises for me is this: is being closer to the maximum give you a good or a bad rating?
The greater the number is, the better.

Secondly, while you do not appear to currently have an overall type rating formula (as in a formula that takes the offensive and defensive type ratings and outputs some number), would such a formula be practical for determining how good a Pokemon's type is overall?
I don't think it's important to do an overall type rating. What would be important is to incorporate the offensive typing rating into the physical and special sweepiness formulae, and the defensive typing rating into the physical and special tankiness formulae. That would ultimately provide a Base Stats & Typing overall Rating.
 
#40
This is a very interesting chart, and I think it will contribute greatly to the CAP process (synergy with BSR). I would be interested to see if the defensive rating could be weighted based on the statistics you collect about how often certain attacking moves are used. This would go some way towards making the defensive ratings reflect the metagame, as well as the distribution of moves out there.
 
#41
One thing I'd like to know is how you came up with offensive type ratings. One problem I see is that dual-types ALWAYS have higher OTR than mono-types, regardless the types in question. This means that something like Normal/Poison performs better than mono-Rock, something I don't think most people would agree with.

For comparison, against the 153 possible types:

Rock
4x effective vs. 6 types
2x effective vs. 44 types
Neutral vs. 67 types
0.5x effective vs. 33 types
0.25x effective vs. 3 types
0x effective vs. 0 types

Normal/Poison (considering only the most effective of the 2 types against a particular type)
4x effective vs. 0 types
2x effective vs. 11 types
Neutral vs. 96 types
0.5x effective vs. 41 types
0.25 effective vs. 4 types
0x effective vs. 1 type

Yes, Normal/Poison has good neutral coverage, but is walled more than Rock, and gets destroyed when it comes to super effective hits.

One potential fix is replacing the value of None with something slightly lower than the value of Normal, so that Rock + None adds to more than Normal + Poison, but Normal + Poison would still add up to more than Normal + None. What that value would be depends entirely on how you got them in the first place.

EDIT: forgot about a bunch of ghost-types, will update shortly, buts it looks even worse for Normal/Poison now
EDIT2: ghost types now accounted for
 
#42
The thing is, the calculations assume that the poke has access to very move, and the bonuses come from having STAB. Being able to hit an opponent with a neutral STAB when your other STAB is ineffective is very nice, and that's why dual-types have higher offensive ratings. The two STABs being walled isn't very important if one has access to other moves as well. The reason Normal/Poison isn't lower is that it can still use Rock-type attacks; Mono-Rock only gets STAB on Rock and a bunch of weaknesses.
 

david stone

Fast-moving, smart, sexy and alarming.
is a Site Staff Alumnusis a Programmer Alumnusis a Super Moderator Alumnusis a Live Chat Contributor Alumnusis a Researcher Alumnusis a Contributor Alumnusis a Battle Server Moderator Alumnus
#43
I disagree with the approach of simply adding up all of the resistances.

Consider a Pokemon neutral to everything, and then consider a Pokemon that is immune to half the types and 2x weak to the other half (with a neutrality to 1 type because there are 17 types). Both will sum to the exact same value (17), but I suspect the "immune to half the types" Pokemon would fare a lot better in a real game.

However, as a Pokemon's overall tankiness increases, the value of resistances is usually overcome by the drawback of weaknesses. In other words, for most Pokemon, the multi-immune Pokemon is better, but if you were to get on the edge case of a super-tanky Pokemon (Blissey's HP with Shuckle's defenses), you probably want to take everything neutral so you don't have a weakness they can hit.
 
#44
The thing is, the calculations assume that the poke has access to very move, and the bonuses come from having STAB. Being able to hit an opponent with a neutral STAB when your other STAB is ineffective is very nice, and that's why dual-types have higher offensive ratings. The two STABs being walled isn't very important if one has access to other moves as well. The reason Normal/Poison isn't lower is that it can still use Rock-type attacks; Mono-Rock only gets STAB on Rock and a bunch of weaknesses.
Rock's weaknesses should have no bearing on its OTR.

Also:

It should be noted that the offensive nature of the typing of a Pokemon shouldn't note movepool at all. For example, just because we have a Fighting-type Pokemon does not mean that it will have Close Combat. Conversely, just because we have a Normal/Flying type Pokemon does not mean that it won't have Close Combat (see: Staraptor). Hence, as I like to repeat over and over again, the offensive nature of a Pokemon is affected the most by its movepool, and not by its typing or stats.
My understanding is the OTR is calculated as though the type in question only has access to its STAB, otherwise this disclaimer would not be there. So no, the Normal/Poison type cannot use Rock moves for the purpose of determining its OTR.

Rock provides better coverage in a single move than Normal and Poison combined. Yes, some types resist Poison that you can hit with Normal, but there are still noticably more types that resist both. Normal/Poisons only advantage is neutral coverage, and not even very impressive neutral coverage. I'd take the far better super effective coverage and harder to wall Rock any day.
 
Status
Not open for further replies.