Kills per match statistic.

[dekzeh]

First of all, sorry if this is not the right section or if this has already been discussed before.


I wonder if it's possible for the Smogon Shoddy Server monthly statistics to include the number of kills per match done by each Pokémon on average.

Often enough we read things such as 'Salamence always gets at least one kill per game' or similar, and having those numbers could be one way to 'quantify' how threatening a Pokemon is.

Perharps this could be done simply by summing the % damage done by each attack and dividing the total per 100 for each match (for each Pokemon obviously). It's just an idea.

 

[Scofield]

Perharps this could be done simply by summing the % damage done by each attack and dividing the total per 100 for each match (for each Pokemon obviously). It's just an idea.

I'm not quite sure if I know what you mean here. I don't think this would work, as each pokemon doesn't run the same set...

Anyways, I have thought about this before, but figured it couldn't work because it's actually quite difficult to define a "kill". If I bullet punch a pokemon at 1 hp, does the bper get the kill or the mon that did 99% to it? Of course, if there were a way to actually calculate this that I can't think of, that would be awesome.

 

[dekzeh]

I think the best would be splitting in two different statistics.

Total damage dealt: This would count the total % damage each pokemon does.
Total kills made: This would count only the number of KO's

Example

Scizor uses U-turn on Blissey and does 55%
later on the match Scizor switches on Blissey again and Uturn it for 50% damage.
on the third time Scizor Superpower's and kills Blissey, doing 100%.

So the total Damage Dealt would be 205%
And Scizor would have one kill.

(Assuming he didn't do any damage to anything else).

 

[Scofield]

Thing is, total % damage dealt is a pretty much useless statistic from what I can gather. What if salamence averages 500% damage? What does that tell you? It kills 5 mons? Maybe it just did 50% damage to blissey with brick break 10 times while blissey pp stalled it to death? Matches that last longer would just yield higher percentages of damage since mons on stall teams can heal themselves more.

Again, with total kills, I feel the numbers can be misinformative. What if blissey was a counter sash version, and scizor comes in to clean up that 1 hp with a nice u turn. Scizor would get credit for the kill, even though salamence did the 99% to bliss?

Now, this idea mostly applies to offensive mons since we're thinking about kills per game. I think we can extend this to defensive mons, by calculating average times blissey uses softboiled per match, how many layers of spikes skarmory gets up on average, etc. I think that would be a very interesting statistic.

 

[zarator]

I think this would be an interesting idea. Of course, sometimes the statistic could lack "interpretation" (like that 1% Blissey kill given to Scizor rather than Salamence), but on average it would still return a significant value - at least, if the Pokémon got enough usage. After all, even if you look at the "probable moveset" of some Pokémon, you get sometimes inaccurate information (heck, Roserade probable moveset is even illegal with Spikes and Sleep Powder on the same set), but this does not mean said stats can be useful nonetheless. So, in short, I totally support this

 

[Garchomp of the Funk]

Maybe this would work if you didn't count damage that was subsequently healed off. Presumably the formula needs to take status conditions and Stealth Rock into account too. Perhaps entry hazard damage should be a statistic independent from any pokémon's damage percent? Maybe a stat for average damage, and a different one for the minimum damage a pokemon is guaranteed to do 75% (or 90%) of the time.

I'm sure X-Act and friends could create some sort of statistic, but any sort of metric like this will inevitably be abused in debates.

 

[X-Act]

I think "average percentage damage dealt by Pokemon" would be a more important and interesting measure than number of kills. As other people already said, if Pokemon A deals 80% damage in one turn, that's more damage than Pokemon B dealing 500% damage in 10 turns.

If it's possible, I'd like to see this.

 

[Scofield]

I think "average percentage damage dealt by Pokemon" would be a more important and interesting measure than number of kills. As other people already said, if Pokemon A deals 80% damage in one turn, that's more damage than Pokemon B dealing 500% damage in 10 turns.

If it's possible, I'd like to see this.

That's a pretty solid idea, probably why you're the stats guy!

Something like, "every time salamence attacks, how much % damage does it do on average (regardless of which attack is used)?" seems like it should be pretty simple to calculate. Now, the problem is in the strict damage clause. Again, if I send out mence to revenge that 1 hp blissey, that could significantly reduce the average. Is there a way to calculate this even for battles with strict damage clause activated? At the same time, you don't want to cheat and count that 300% mence did to forretress with fire blast, so a cap of 100% should be made for each move it does.....or should it?

 

[zorbees]

Yes, but that only takes offensive pokemon into account, X-Act. Think of Blissey. Blissey will only be doing 100 damage per turn (assuming it has Seismic Toss like many do) but it can sponge many hits easily, allowing it to use Seismic Toss more often. Imagine Blissey against an offensive Suicune (meaning no rest). Blissey will defeat Suicune (assuming no switching), but will only do 100 damage per turn. Mixmence, on the other hand, can do over 200 damage (with two different moves, no less), but one on one, loses against Suicune. (Not the most practical of situations, but whatever). What I think may be good is something you were talking about in the Portrait of an Uber thread, the ratio of the amount of damage it can deal with the amount of damage it can withstand, unfortunately, you said you'll probably never be able to do that mathematically and objectively, so yeah...

Anyways, if you could figure it out, it would be very interesting, as you said (and I assume you still believe) that it is a measure of how good a pokemon is.

 

[X-Act]

Yes, but that only takes offensive pokemon into account, X-Act. Think of Blissey. Blissey will only be doing 100 damage per turn (assuming it has Seismic Toss like many do) but it can sponge many hits easily, allowing it to use Seismic Toss more often. Imagine Blissey against an offensive Suicune (meaning no rest). Blissey will defeat Suicune (assuming no switching), but will only do 100 damage per turn. Mixmence, on the other hand, can do over 200 damage (with two different moves, no less), but one on one, loses against Suicune. (Not the most practical of situations, but whatever). What I think may be good is something you were talking about in the Portrait of an Uber thread, the ratio of the amount of damage it can deal with the amount of damage it can withstand, unfortunately, you said you'll probably never be able to do that mathematically and objectively, so yeah...

Anyways, if you could figure it out, it would be very interesting, as you said (and I assume you still believe) that it is a measure of how good a pokemon is.

I never said that this statistic directly correlates with how good or bad a Pokemon is. Yes, it only takes care of the offensive nature of the Pokemon. So what?

Also, to scofield, strict damage clause doesn't matter in the long run. After all, it would be applied to every Pokemon, not just the Salamence in your example.

 

[zorbees]

I never said that this statistic directly correlates with how good or bad a Pokemon is. Yes, it only takes care of the offensive nature of the Pokemon. So what?

Meh, I don't really feel like arguing, especially since you said that my favored statistic is implausible anyways.

 

[Staraptor Call]

Yes, but that only takes offensive pokemon into account, X-Act. Think of Blissey. Blissey will only be doing 100 damage per turn (assuming it has Seismic Toss like many do) but it can sponge many hits easily, allowing it to use Seismic Toss more often. Imagine Blissey against an offensive Suicune (meaning no rest). Blissey will defeat Suicune (assuming no switching), but will only do 100 damage per turn. Mixmence, on the other hand, can do over 200 damage (with two different moves, no less), but one on one, loses against Suicune. (Not the most practical of situations, but whatever). What I think may be good is something you were talking about in the Portrait of an Uber thread, the ratio of the amount of damage it can deal with the amount of damage it can withstand, unfortunately, you said you'll probably never be able to do that mathematically and objectively, so yeah...

Anyways, if you could figure it out, it would be very interesting, as you said (and I assume you still believe) that it is a measure of how good a pokemon is.

Could Skarmpoints and Blisspoints be used to measure the amount of damage a Pokemon can withstand?

 

[Upside]

Couldn't we do something along the lines of average damage per turn?

A very simple formula could be...

(Damage)/(# of attacks)

This would give a nice estimate of how much of an offensive threat the rated pokemon is. Actually, we could probably just do 2 or 3 different statistics to quantify how much of a threat a pokemon is. Damage per turn, total damage over the course of the match, and number of kills. While each has it's very obvious flaws, the combination of all 3 would give a good general idea of how much of a threat that pokemon is. It definitely wouldn't be perfect, but in a game as in-depth as pokemon, our quantifications are never going to be perfect.

 

[dekzeh]

I actually like the idea of average damage per attack, as long as we also have the average number of attacks per match.

The main reason is that I'd rather use something that does 150 damage per turn, but attacks 10 times during a match instead of something with 500 damage per turn that rarely attacks more then once.