Jump to content
RichardNixon

WN9 candidate prototype

Recommended Posts

2 hours ago, RichardNixon said:

Account WN9? You could argue that it wasn't universal anyway, as it only uses random battles. The choice is whether to generate a really bad number for the handful of arty-only players (6 on EU), or no number at all. They can still use recent WN9, which will give them a sensible result.

I was more referring to the degree of tier influence manipulation.  I can't use this metric anymore as it stands.  It's completely worthless as a gauge of skill in any area of the game I would want to touch.  I can't even pair it with a second metric or data point to discern a meaningful difference between players in crowds of similar numbers to get the information I want.  WN8 could be paired with looking at average tier if a high tier player wanted to spot a mid or low tier player, but there's nothing to pair with WN9 if I want to find highly rated low tier players.  Per tank battles is about as close as I could get, but the weighting is so strong that high battle counts stop mattering if your subject steps even one tier higher than what's relevant, so I may as well be using a whole different metric at that point.

Share this post


Link to post
Share on other sites
27 minutes ago, Kuroialty said:

I was more referring to the degree of tier influence manipulation.

Oh, that. The trouble is that the introduction of newbie MM completely trashed the tier 1-3 data, rather like 8.6 for arty. On top of that, WG literally halved the capability of the most popular tier 2. As you can no longer tell from historical data whether a player is good at low tiers or just played at a good time, the brutal weight caps on account WN9 make sense. Otherwise I'd have used flat weights from 1-4.

Unlike arty, even the recents don't really work for tier 1-3 because the newbie-MM effect varies with server & battle count. They do still exist, but I had to set the expected values at pre-NMM levels.

Share this post


Link to post
Share on other sites
11 minutes ago, RichardNixon said:

Oh, that. The trouble is that the introduction of newbie MM completely trashed the tier 1-3 data, rather like 8.6 for arty. On top of that, WG literally halved the capability of the most popular tier 2. As you can no longer tell from historical data whether a player is good at low tiers or just played at a good time, the brutal weight caps on account WN9 make sense. Otherwise I'd have used flat weights from 1-4.

Unlike arty, even the recents don't really work for tier 1-3 because the newbie-MM effect varies with server & battle count. They do still exist, but I had to set the expected values at pre-NMM levels.

That feels like a much more valid reason for the steps taken than what I've seen in the thread more recently.

You mention in the FAQ that adding assisted damage and associated battle count would help improve the metric by allowing that data to be analyzed and possibly used.  Would a battle count since the implementation of newbie MM be helpful at all in gauging tier 1-3 performance?

Share this post


Link to post
Share on other sites
17 minutes ago, Kuroialty said:

You mention in the FAQ that adding assisted damage and associated battle count would help improve the metric by allowing that data to be analyzed and possibly used.  Would a battle count since the implementation of newbie MM be helpful at all in gauging tier 1-3 performance?

Partially. There are actually three cases with very different expected values: Played before NMM, played after NMM with <2500 battles, and played after NMM with 2500+ battles. Before & after NMM could be derived from a database snapshot (assuming WG have those), but the 2500 battle mark would need active updates.

Share this post


Link to post
Share on other sites
4 hours ago, Trobs said:

Account WN9 - N/A

First column should be N/A, because it's not applicable. If the second column (tanks/stats) doesn't have a number then you may need shift/ctrl-refresh. Web caching sucks. You're at 914.5 anyway.

Other news:

- KR should be added to the experimental stats page soon, because WG suddenly made the application IDs universal. They also made the WoT console API public, although that would take rather more work to implement.

- Data for the next expected values pass should be ready on Monday 1st August. I'll probably call that 1.0 and then release WN9 generally, which basically means paging in Never/Phalynx/Seriych.

Share this post


Link to post
Share on other sites

So it turns out that the Grille totally fucked up tier 10. I need to rework the normalization system so that the expected values don't lurch around too wildly. Might take a couple of days.

 

3 hours ago, Stige said:

Blue WN9

I don't bother working around the account reset bug, as then I'd have to document the method for other sites. You could try begging WG to fix their shit.

I suppose I could add a probable-reset warning.

Share this post


Link to post
Share on other sites
2 hours ago, RichardNixon said:

So it turns out that the Grille totally fucked up tier 10. I need to rework the normalization system so that the expected values don't lurch around too wildly. Might take a couple of days.

 

I don't bother working around the account reset bug, as then I'd have to document the method for other sites. You could try begging WG to fix their shit.

I suppose I could add a probable-reset warning.

Do what needs to be done. A couple more days of waiting won't change anything. :)

Would you be so kind, to briefly explain what did the grille change statistically?

Share this post


Link to post
Share on other sites
4 hours ago, TAdoo87 said:

Would you be so kind, to briefly explain what did the grille change statistically?

Main disruptive features of the Grille:

  1. Easily the highest damage output at tier 10 for most players.
  2. Doesn't have much HP itself.
  3. Extremely popular.

Together, those three points drag down the damage output of all other tier 8-10 tanks, especially tier 10s. The available HP is lower, and the Grille's taking more of it.

Share this post


Link to post
Share on other sites
10 hours ago, RichardNixon said:

Main disruptive features of the Grille:

  1. Easily the highest damage output at tier 10 for most players.
  2. Doesn't have much HP itself.
  3. Extremely popular.

Together, those three points drag down the damage output of all other tier 8-10 tanks, especially tier 10s. The available HP is lower, and the Grille's taking more of it.

Glad you finally noticed it, add 50/51 to that as well then, you will prob. see similar results, also not as bad as from Grille I guess, but prob. still way more disruptive as compared to any other t10 medium.

There are a lot of discussions from high skill players/streamers in the last few month, which I personally agree with, that it was way easier to farm dmg in t10s in 2012-2014, than it is now. And that you can't compare DpG anymore, without knowing, when the games were played. Atm., it's mainly caused by grilles and 50/51. They just take a lot of clean up dmg late game as most extreme negative impact example, where you get 1-2 320/390 alpha shots in, if you manage to get in time, but they get 1-2 750 alpha ones or 1-2 clips. And that's only one of many aspects IMO.

Straik for example said on stream, that he has hard time keeping over 4k dpg in t10s without try harding and is somewhere around 4.5k on try hard mode, where as back in 2012-2014 he would keep 4.5k dpg with the same effort and that even players like iMeeGo, who could get ~ 5k dpg hard core try harding are now barely getting 4.5k dpg, even though latest example might be also caused due to different (lower) motivational aspects, not only current state of the game.

Would it be possible to check few top notch players from ru/eu/na and compare their DpG results in 2-3 of their t10 meds/heavies against different WOT patches maybe?

Share this post


Link to post
Share on other sites
On 8/1/2016 at 3:55 PM, GrimmReaperBG said:

I hope it will be any time soon, but not next year :)

On 8/2/2016 at 0:40 PM, GHOSTK1LL3R said:

So whats going to happen? IS this going to replace WN8 on wotlabs and be implemented to the official XVM?

"When it's done, it's done." - SerB

It will be soon, don't worry guys, but this takes a lot of work.

Share this post


Link to post
Share on other sites
6 hours ago, lazydot said:

Glad you finally noticed it, add 50/51 to that as well then, you will prob. see similar results, also not as bad as from Grille I guess, but prob. still way more disruptive as compared to any other t10 medium.

The 50/51 doesn't have the popularity (yet) or the damage output for mediocre players. That effect was absorbable, even together with the M48 & E5 overbuffs, partly because gradual changes are more acceptable than sudden ones, but also because the Grille had a much larger effect.

 

6 hours ago, lazydot said:

Straik for example said on stream, that he has hard time keeping over 4k dpg in t10s without try harding and is somewhere around 4.5k on try hard mode, where as back in 2012-2014 he would keep 4.5k dpg with the same effort and that even players like iMeeGo, who could get ~ 5k dpg hard core try harding are now barely getting 4.5k dpg, even though latest example might be also caused due to different (lower) motivational aspects, not only current state of the game.

The Grille-spam's more like a 100-150 damage penalty at that level, assuming that you don't play it yourself. I suspect there are stronger historical effects, notably an increase in arty counts since personal missions, map changes, and player-base stagnation leading to higher average skill & tier distribution.

There is an option of setting the scales according to actual player distributions, but this only makes any sense if you set them separately for each server, would need a huge amount of additional update work (the current WN9 method is bad enough), and may lurch around with seasonal changes. There's also no guarantee that the top 1% of players on each server are equally skilled.

The alternative is to minimise numerical jumps on updates, and accept that the unicum mark varies by server, and may become more or less difficult over time. I'm not sure anything else is practical. I can publish distribution tables occasionally, but I don't think they can be part of the update process.

 

8 hours ago, lazydot said:

Would it be possible to check few top notch players from ru/eu/na and compare their DpG results in 2-3 of their t10 meds/heavies against different WOT patches maybe?

My data doesn't go very far back. The best I can do is older EU accounts from late 2014. More general samples from mid 2015 at the earliest.

 

Share this post


Link to post
Share on other sites

Ok, changed my mind. There's a relatively straightforward way to implement per-server and per-metric distribution scaling, mostly with data that I'm already collecting. I'll throw something together and see what it looks like.

Share this post


Link to post
Share on other sites
5 hours ago, RichardNixon said:

Ok, changed my mind. There's a relatively straightforward way to implement per-server and per-metric distribution scaling, mostly with data that I'm already collecting. I'll throw something together and see what it looks like.

Interesting, can you go into more detail?

Share this post


Link to post
Share on other sites
11 hours ago, RichardNixon said:

 I suspect there are stronger historical effects, notably an increase in arty counts since personal missions, map changes, and player-base stagnation leading to higher average skill & tier distribution.

What about light tanks?  It seems like there are usually more light tanks in queue than any other class by a huge margin since the missions started and with all the new light tanks in the tech trees.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...