Jump to content
Garbad

xTE and the obsolescence of Wn8

Who is more of a retard?  

196 members have voted

  1. 1. Who is more of a retard?



Recommended Posts

> since that's where the fake unicums came from,where they get ppl to believe they are better than 99.9% of the players,

which they are not.

 

But statistical data for entire populations says you need to be better than said 99.9% to get unicum and 99.99% to get supercum.

So you statement isn't very logical, don't you think?

 

No one said that all players with unicums stats are any good, if you look at someone's stats, you 1st you check his number of battles, his WR, then his WN8/WNx, then you go to his top tiers/scouts and check dmg/frags/spots then you have a pretty good picture (also telling you which tanks the player mostly likely started with before he got to a level where he is at currently, like 62a vs 140 comparison)

 

>COMPARING TO OTHER PLAYERS IN A CERTAIN TANK.

Its comparing your performance to the the few best of the best and not to other players as far as I understand that.

 

After all I'm not disillusioned about wn8, but its a good starting point and in the most cases I know what I can expect from 2.5k wn8 player and then 3k, 3.5k, but even then you meet green/blue bobs, who just took a very long learning curve or just started to care more about getting better recently and have exceptional values in few tanks they researched recently also overall picture is kinda mediocre.

 

I personally think that noobmeter's and then wn8 ratings are pretty much the best in this order and xte as it is now is not good enough (WR+spots correlation to name 2 major points), even though from statistical knowledge point of view I have no remote clue on the subject.

Link to post
Share on other sites

The thing is, the scale is something that is way off

 

If you have exactly the average damage and frags it gives you something like 30 which is way too low for what is supposedly "average"

 

My best tanks where I did way better than average are still 50-60ish

 

So IMO the scale they chose is completely unintuitive, it maybe works for that 0.1mm difference in epeen length but makes most people look worse than they are

Link to post
Share on other sites

So is your issue that it could hurt the feelings of non-unica?  Honestly, that seems to be what's behind some of your posts.  Personally, I find it refreshing that xTE seems to rate the player more in line with the overall community; top and bottom. 

 

You'll never see or hear me refer to myself even as a "recent unicum" because aside from a few specific tanks, I don't feel like my play is anywhere near the number in my recent stats.  I feel more in line with my teal/blue overall and xTE reiterates that.  Criticism is useful for being able to grow.  Something like xTE enables players to see what's near the ceiling in a tank and ask themselves where they can improve to get there. 

 

As far as your point that WN8 is applicable to sub-unica which makes up the majority of players: I agree in general with you.  The issue I have with that is that the majority of players don't care about any rating or improving their gameplay.  Therefore, why create a metric that benefits people who don't care about it?  That's just a waste of resources.  Creating a metric for a smaller population, the majority of which care at least enough to have a debate about its efficacy, goes a lot further to spurring helpful dialogue on these forums and advancing the skills of those who actually give a damn.\

 

Edit:  Something else occurred to me as I thought about it.  I guess what I like about xTE vs WN8 is that WN8 tells you how you're doing in that tank compared to unica, bots, "I just play for fun" 44%ers and everything whereas xTE tells you how you're doing in that tank compared to it's real world ceiling.  I can get super-uni wn8 in some tanks.  That dark purple can make you feel good.  But that doesn't give me any information whatsoever about what others are doing even beyond that threshold. 

Link to post
Share on other sites

xtc works perfect to compare people tank vs tank, say how i do in E50 and how some1 else does in E50, it also works good when comparing my E50 play with my M46 play, or with somebody else hes m46 play

 

its for in-tier-comparisons, a superb metric, untill people start farming damage, so that means it will loose its value very fast....

Link to post
Share on other sites

xtc works perfect to compare people tank vs tank, say how i do in E50 and how some1 else does in E50, it also works good when comparing my E50 play with my M46 play, or with somebody else hes m46 play

 

its for in-tier-comparisons, a superb metric, untill people start farming damage, so that means it will loose its value very fast....

No more so than WN8.  And that will always be the issue with any statistical metric.  Once people figure out how it works, the numbers can be gamed if you so choose.

Link to post
Share on other sites

Is there actually a possibility of somehow using this accuracy up high that xTC has in wn8?

 

I know you guys are trying a system for recency, or at least I`m under the impression of this, but this really high level high level accuracy is nice and would be good to adopt/modify if the metric stays relevant lower down. (Which I have the impression xTC does not)

 

You are probably already working on this, if so my apologies.

 

Expected values based on recent data give you good high end accuracy if used correctly. Probably better than xTC. The trouble is that everyone on WoTLabs has their own ideas of what expected values should be for different tanks, and they approach the subject aggressively and with deeply-engrained bias. The work's basically done but I don't have the stomach for the fight.

 

There's also the issue of the lack of usable assisted damage. Without that, there are few improvements for the vast majority of players, and the playstyle bias problems are impossible to resolve. You can fix the problems at the upper end of the skill range and improve per-tank ratings, but that's about it. Motivation for new metrics would be a lot stronger if they had something for everyone.

Link to post
Share on other sites
Second,why should you eliminate bots if they're doing fine,like a normal (bad) player does,wouldn't that make all those statistics inaccurate?

 

Well no, that was what I pointed out initially. Since WN8 is meant to measure the skill of a human player, whether a bot does better than a human is irrelevant. The statistic is meant to measure human players, not coding. If a bot was a unicum, I'd still think it'd be a clutter stat, since it isn't a player that's being evaluated, it's a bot. It'd be kind of like asking for the opinion of males in a survey specifically directed towards the opinion of females. It's not the goal of the survey, therefore the data is meaningless.

Link to post
Share on other sites

Since WN8 is meant to measure the skill of a human player, whether a bot does better than a human is irrelevant. The statistic is meant to measure human players, not coding. If a bot was a unicum, I'd still think it'd be a clutter stat, since it isn't a player that's being evaluated, it's a bot. It'd be kind of like asking for the opinion of males in a survey specifically directed towards the opinion of females. It's not the goal of the survey, therefore the data is meaningless.

 

Uhhh.  There is nothing about the rating specific to humans, just a measure of effectiveness in battle.  It's just as valid if it's a bot or a human, it's simply a measure.

Link to post
Share on other sites

wn8 seems to work fairly well to a certain point I think after looking into it a little bit. A green playing lights won't suddenly get purple stats. The only issue people have is with the bragging rights for post 3k wn8. I retract my previous statements in this thread...

Link to post
Share on other sites

So why is everyone calling this xTC if it's actually named xTE?

 

Maybe Garbad was originally testing who bothered to read the post. After reading the damned thread title 9000 times it starts to burn a hole in your brain.

Link to post
Share on other sites

Thread title fixed

 

Might as well change the date to 2013 as well, not sure why the OP waited 2 years to copy and paste the info on TEFF/xTE here and make like it was news.

 

The discussion between the OP and others from the official forum from 2013 is here

Link to post
Share on other sites

So when looking at xTC im assuming a 50 means you are at the average damage per game.  Yet when I pull up my type 59 it shows I am at 46 yet my dpg is well above the average, what is the reason for this, or am I missing something?

 

 

shot_004.jpg

Edited by xMakav3l1x
Link to post
Share on other sites

So when looking at xTC im assuming a 50 means you are at the average damage per game.  Yet when I pull up my type 59 it shows I am at 46 yet my dpg is well above the average, what is the reason for this, or am I missing something?

1) Its 'xTE' not 'xTC' - the OP cant even cut and paste

2) Average on the 'xvm' scale is 29, a score of 50 puts you at about the 80th percentile. No, I have no idea why xvm thought such a scale would be useful.

Link to post
Share on other sites

The xTE scale looks like this:

xTE% Pop
00
1110
1720
2130
2540
2950
33.560
3870
4480
52.590
6095
75.599
81.599.5
92.599.9
XX100

There's more room at the top to separate the "mere" unicums from the super unicums. 

You can find a color chart by googling images for  "wot xTE color scale."  It should be one of the first images.

Link to post
Share on other sites

Such a logical scale.

Plus its fucked by recency bias, IE me being XX or close in the 140 despite it probably being 4000dpg if it was based of recents (I am like 3580~)

Link to post
Share on other sites

Or do one great game out of the 2 or 3 I played in pubs with my SP. All of a sudden super uni xTE. Kek samplesize. Though I suppose that is an issue for any of the metrics.

Edited by KenadianCSJ
Link to post
Share on other sites

Wow, actually went to a per vehicle rating. Considering it is per vehicle why not include win rate (with some max to mitigate platoon effect) and spotting damage?

Link to post
Share on other sites

Wow, actually went to a per vehicle rating. Considering it is per vehicle why not include win rate (with some max to mitigate platoon effect) and spotting damage?

For a start, winrate's generally too inaccurate to be useful per-tank. Secondly, if you have a maximum to mitigate the platoon padding, it'll be applied on almost every tank with their method, so there would be no real per-tank or even per-tier adjustment. There are methods to do tank-adjusted winrate, but xTE isn't going to work because of its dependence on the maximum.

Spotting damage is interesting because it should be accessible in-client. Ideally you'd use it for per-game stat mods at least, rather than those horrible per-game WN8 ratings. I'm not sure whether xTE is reading client stats or pure API.

 

Link to post
Share on other sites

Wow, actually went to a per vehicle rating. Considering it is per vehicle why not include win rate (with some max to mitigate platoon effect) and spotting damage?

For a start, winrate's generally too inaccurate to be useful per-tank. Secondly, if you have a maximum to mitigate the platoon padding, it'll be applied on almost every tank with their method, so there would be no real per-tank or even per-tier adjustment. There are methods to do tank-adjusted winrate, but xTE isn't going to work because of its dependence on the maximum.

Spotting damage is interesting because it should be accessible in-client. Ideally you'd use it for per-game stat mods at least, rather than those horrible per-game WN8 ratings. I'm not sure whether xTE is reading client stats or pure API.

 

Not really here to pick a fight (retired from tanks) but how is win-rate too inaccurate (whatever you mean by that) to be useful per-tank but damage and kills doesn't have an issue?

Link to post
Share on other sites

Not really here to pick a fight (retired from tanks) but how is win-rate too inaccurate (whatever you mean by that) to be useful per-tank but damage and kills doesn't have an issue?

 

They all have issues, but essentially the standard deviation of winrate is far higher than that for damage, relative to the difference between an average player and an AFKer. Winrate needs at least 10x as many games as damage to get the same signal to noise ratio.

IIRC kills and spots are closer to winrate for inaccuracy. Defence is worse.

 

Link to post
Share on other sites

Wow, actually went to a per vehicle rating. Considering it is per vehicle why not include win rate (with some max to mitigate platoon effect) and spotting damage?

For a start, winrate's generally too inaccurate to be useful per-tank. Secondly, if you have a maximum to mitigate the platoon padding, it'll be applied on almost every tank with their method, so there would be no real per-tank or even per-tier adjustment. There are methods to do tank-adjusted winrate, but xTE isn't going to work because of its dependence on the maximum.

Spotting damage is interesting because it should be accessible in-client. Ideally you'd use it for per-game stat mods at least, rather than those horrible per-game WN8 ratings. I'm not sure whether xTE is reading client stats or pure API.

 

Not really here to pick a fight (retired from tanks) but how is win-rate too inaccurate (whatever you mean by that) to be useful per-tank but damage and kills doesn't have an issue?

Because you can easily have 45% winrate while playing around 3K WN8 for 100 games and more. It happens.

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...