Official Smogon University Usage Statistics Discussion Thread, mk.2

Status
Not open for further replies.
mallpal, that came up in the basculin discussion and is part of the reason why we decided to tier the basculins together (also tiering all the cosplay pikachus together)
 

Arcticblast

Trans rights are human rights
is a Forum Moderatoris a Tiering Contributoris a Social Media Contributor Alumnusis a Senior Staff Member Alumnusis a Community Contributor Alumnusis a Battle Simulator Moderator Alumnusis a Past SPL Champion
hi, sorry if this is the wrong place to ask this, but shouldn't keldeo and keldeo-resolute be tiered separately? if we're following the precedent set in-game, there's no way for base-form keld to know secret sword as far as i'm aware (i know that's not how it's implemented on ps! but that's easy enough to change, i presume), and this is a significant enough difference to merit separating them imo.

i'm asking this on the back of the basculin case, obviously, because it's an analogous case.
You can actually trade a Secret Sword Keldeo to Black/White from BW2, where it'll revert to normal form but keep the move, and then transfer it to XY without it changing form.
 
I agree with MarioWithLasers. As to your solution, while doing graphs would definitely allow for a more complete display of common stats individually, I think it would be worse overall since there's likely no way to show more possibilities without complitcating it too much (for instance, how often does Landorus-T run 88 neutral speed? How about 168 positive?). While it would surely be nice to have alongside the spreads as listed now, I don't think it would be good as a substitution. As to drawing the line, that can be adressed later, as points can be made for both sides in some cases (he naive x hasty thing, for example, has a significant impact on damage calcs); I don't think anyone would disagree with combining two spreads that only differ in 4 Def x 4 SpDef EVs.
What happened to this suggestion ? Forgotten ? I actually agree with this, and I'll be further with this :
Rotom-Wash spreads said:
Spreads |
| Bold:248/0/216/0/0/44 13.633% |
| Calm:252/0/0/4/252/0 12.826% |
| Bold:252/0/212/0/0/44 10.206% |
| Bold:252/0/252/4/0/0 8.139% |
| Timid:0/0/0/252/4/252 4.235% |
| Calm:252/0/0/0/212/44 2.757% |
| Other 48.203%|
Now, if a player wants to know whether the most common set he will face is a physically defensive or a specially defensive set, he may think this the the physical one because 13.6+10.2+8.1 = 32% while 12.8+2.7=15.5%. But in fact, those 48% left can be all weird specially defensive sets. And those one doesn't have any 4 EV difference (which could be ignored when showing set as previoulsy said) but sometimes 44 EVs difference. So may be the indication of "Specially defensive" / "Physically defensive" / "Mixed Wall" / "Fast physical attacker" / "Bulky physical attacker" / "Bulky Fast pivot" or any combination of those things (even for instance "Specially bulky physical atacker" for a 252 Atk/252SpDef set) may be better.
 
Ramplestilskin, that's actually a very elegant solution. Probably the best way to present the information is to categorize them based on stat spread (as opposed to EV spread-- a 252 HP / 252+ Def Deoxys-A is still going to be more offensive than bulky), then also provide a table of Stat and % running: Max + Investment, Max Neutral Investment, No Investment Neutral, No - Investment, Other. I'll put it on my to-do list.
 
With the combining of the VGC 2015 and Battlespot Doubles ladder what, if any, effect will that have on this months stats for the "bring 6 choose 4 double battles at level 50" ladder
 
So like, will the numbers from both defunct ladders from the past 27 days get added together in this month's stats for the new ladder?
 
Can you merge "Battle Spot Doubles", "VGC 2015" and "Battle Spot Doubles (VGC 2015)" into one format, only for January? My suggestion is merge it into: VGC 2015.
 
I read that but either I didn't understand or it seemed pretty strange to me. referring to this :
Antar said:
  • A brand new player, just starting out, whose rating is 1500±350 will be weighted 0.5.
  • I have a mediocre OU team i sometimes play on PS under an alt. It currently has a provisional rating of 1576±105. Its weighting is 0.77.
  • The person who I just demolished with that team (you have to be pretty bad...) has a provisional rating of 1394±139. Weighting is 0.223
  • My good (not great) OU team has a rating of 1946±177. Weighting is 0.994
  • The person at the top of the ladder right now has a rating of 2120±55. Their weighting is 1.0, for all intents and purposes.

and seeing as, for instance, OU avg weight / team is 0.062 (which means a really bad level of play), no type in monotype metagame goes further than 0.02, and the fact that this weight is about a "player weight" and not a "team weight", I think I miss something ><
 
I read that but either I didn't understand or it seemed pretty strange to me. referring to this :
and seeing as, for instance, OU avg weight / team is 0.062 (which means a really bad level of play), no type in monotype metagame goes further than 0.02, and the fact that this weight is about a "player weight" and not a "team weight", I think I miss something ><
That's because you're looking at ou-1695. 1695 glicko is very high and - to put it simple, only the best players ("top 6.2%") contribute to these stats. In ou-1500 avg weight is much closer to 50%. Monotype I assume it's because most teams do not fit into the type you're looking at so are weighted 0, which explains the low numbers.

Antar, do some pokemon entries not get listed in the JSON files? Each time I sum up the pokemon abilities of ou-1500.json then divide by the sum in ou-0, I get a result that's 2~5% off the avg. weight/team number. Specifically it's always below the average weight (so I can't assume the gap difference is from players having <6 mons) and the same thing also happens for 1695 & 1825. So I'm wondering if perhaps some pokemon get filtered out because their usage is too low?

Also it'd be nice if avg weight had more decimal places.. like 6 or 7 instead of just 3. Since "cutoff deviation" in the JSONs don;t work....
 
Last edited:
QxC4eva Might be explained by teams with less than six Pokemon. And you're correct about monotype. What do you mean "cutoff deviation" doesn't work?
 
QxC4eva Might be explained by teams with less than six Pokemon. And you're correct about monotype. What do you mean "cutoff deviation" doesn't work?
It can't be. That would suggest that players using less pokemon have higher ratings. Still, I'll take that as a confirmation that no entries get filtered out.

There's a field called "cutoff deviation" under the info header of each json. I've not seen one where the value isn't 0, so chances are it's not working. I think it's meant to show which standard deviation of the player base is being displayed (so running CDF() on it gives the average weight per team) but thinking again I might be wrong. What's it for anyway?
 
It can't be. That would suggest that players using less pokemon have higher ratings. Still, I'll take that as a confirmation that no entries get filtered out.
Okay, I'm remembering now--only Pokemon that get included in the moveset stats get into the json.

There's a field called "cutoff deviation" under the info header of each json. I've not seen one where the value isn't 0, so chances are it's not working. I think it's meant to show which standard deviation of the player base is being displayed (so running CDF() on it gives the average weight per team) but thinking again I might be wrong. What's it for anyway?
Cutoff deviation is just something that's not being used. It was an idea I toyed around with at some point, but I'm not using it, and the field is just there.
 

PISTOLERO

I come to bury Caesar, not to praise him.
Thank you for continually compiling usage statistics Antar, I've been doing the SPL Ubers statistics and I realised how much work it is, it's appreciated
 
Does anyone know what the values in the chaos.json files mean? I'll give you an example from November, ou, 0 ranking:

Code:
{
  "Fletchinder": {
    "Abilities": {
      "flamebody": 32.062900673043,
      "galewings": 97.777307721119
    },
    ...
    "Raw count": 426,
    ...
}
I'd initially assumed that the ability values would be integers, that would tell me how many Fletchinder had Flame Body and how many had Gale Wings (and perhaps some that couldn't be inferred!). This clearly can't be the case. Then my brain moved on to percentages, again, clearly cannot be the case.

Does anyone know what these values represent, and if that's the same across Items, Spreads, Moves, "Checks and Counters" and Team Mates?
 
Imran, the values are the sum of the weights.
But shouldn't everything be a whole number for OU-0?

Also does (Total battles) * (Avg. weight/team) * 12 give the total number of pokemon used?

OU-0.txt:
Total battles = 1292333
Avg. weight/team = 1.0
Landorus-T usage = 16.94389%

OU-0.json:
Landorus-T usage = {"intimidate": 555420.0}

(555420.0 * 6) / (1292333 * 1.0 * 12) = 21.48904%

Both values are different...?
 
Status
Not open for further replies.

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

Top