Jump to content

good idea?  

11 members have voted

  1. 1. Would this display player skill more accurately than a 1K battle recent?



Recommended Posts

It's been known for a while that while recents display definite changes in player skill, they're not perfect because of a variety of factors. For 60 day stats (the original recents standard), it could be someone playing 1 really good game every 60 days or otherwise leading to extremely low sample sizes.

Having 1K battle recents fixed the sample sizes issue, but left unaddressed other problems such as metrics-padding and the influence of habits, as well as sample size spread across various tanks, meaning that recents became close to meaningless in the cases of most teals and higher who had done a little reading about WN8 and learned to platoon (case in point: my recent. I'm really a blue recent, I just play more Leo PTA than what's healthy for me.)

Thus, my proposal is to make recents more reflective of individuals by displaying stats for a portion of their overall number of battles (say, 30% or 40%). I know I'm going to get flak from those who have low overalls with high recents and large numbers of battles for this, but the reasoning is that play really is mostly instinctive for the vast majority of players and habit-based. It's accepted that a dark blue overall with light purple recents is likely better than a teal overall with deep purple recents, discounting all other factors, on the basis that they have more good gameplay habits (such as mechanical skills and quick judgement of the minimap) ingrained, meaning that they win more solo (the elusive de-facto definition of skill).

This percentage-of-total-battles based recent also largely fixes the issue of sample size across various tanks, because it reflects better how much a player has acclimated to playing a particular tank or class, and in the case of most players around here, 30% or 40% of their total battles is well over 1000 battles. If someone could run the math on at what the highest percentage of battles is among players with 1K battle recents at least 1 color above their overall where there's a significant change in their statistics, that would be great for determining a fixed point.

Habit formation is almost the same in humans as in mice and the same among ourselves, as an article from Scientific American (links to studies in article) shows, so it would be reasonable to assume that there's a defined point in which new habits in game-play begin to supersede old ones, based on how ingrained they are in our psyche. There's a reason why the best players are those like Kewei and CarbonWard, who learned from the start by soloing in harsh environments (high Tiers just after release); by happenstance or  deliberate choice, they combined a will to improve with good game-play from the start with what's mechanically the most punishing gameplay, without the flawed and rng-based tactics present at low Tiers, owing to the lenient gameplay and levels of player skill. The opposite can be seen in seal-clubbers trying to play high Tiers, because it's literally impossible to fully kick habits (you can only suppress them). I hypothesize that the more battles someone has at a low skill level, the harder it is for them to suppress their effects.

Obviously, this doesn't fix WN8's issues with padding in certain tanks, but it can at least help to alleviate certain unwarranted statistical outliers in recents, and more accurately display someone's abilities at a glance. The service record will never be beaten for full-on evaluations, but progress is progress I suppose.

PS: I don't know if this is the best forum for this kind of discussion, but it's the best I could think of. Feel free to move this, mods.

*runs away from BULBA riot*

Link to post
Share on other sites

Not sure about the about habit, but assuming all other variables are the same, a green with deep purple(3500) is easily better than a teal with just purple recent(2500) in both consistency and map awareness.

Link to post
Share on other sites

i think the benchmarks for blue/uni/super uni would change as most peoples recents are higher than their overall, thus increasing the average skill level. so to maintain the 1%/.1%/.01% server population ratio the wn8 benchmark would increase. i would be curious of the actual values 

Link to post
Share on other sites

Last 30-40% of all overall battles would cover about two years of games for me. That's not really recent, is it?

Your recent would count battles as of April 16, 2014, which was 5555 battles ago for you (a little over 30% of your overall) or ~1.33 years, which isn't a terrible large time change. That's not to mention that it includes everything from there to now, giving a more accurate sample size when factors such as tank sample sizes are taken into account. The last really big changes to the game came in late summer 2013 (think patches 8.4-8.6), so it's still fairly reflective. Maps have changed, yes, but what habits worked then probably work now. A lot of people well-versed in how metrics work thought Garbad's idea of a 6K battle recent was good as well. That's even worse on your behalf.

Not sure about the about habit, but assuming all other variables are the same, a green with deep purple(3500) is easily better than a teal with just purple recent(2500) in both consistency and map awareness.

That's pretty loaded. Does the teal drive stuff like a VK 4502B and the green an E50? Do they win the same amount solo (around 58% I would guess for both)? Is there some other factor massively influencing it?

 

Link to post
Share on other sites

That's pretty loaded. Does the teal drive stuff like a VK 4502B and the green an E50? Do they win the same amount solo (around 58% I would guess for both)? Is there some other factor massively influencing it? 

By assuming all variables are the same it is assumed they are both driving the same tank.

Link to post
Share on other sites

I don't think it's possible for a green to get 3500 recent all of the sudden without having some major padding tanks (which is MaxL_1023 levels of play), which would more than certainly lift a teal up as well if it involves padding tanks. Like I said, you and Kewei are proof of that concept. Also sorry for being bad@reading. Plus, the changes in habits would be reflected over time if they can truly maintain that level of play.

 

And yes Med, I'm a shitter. That doesn't necessarily invalidate my point in making this topic though, does it?

Link to post
Share on other sites

Well, there's Dia, who had double green overall and deep purple 2 years ago, plus there are plenty of blues with 4000/70+ these days who i would rate higher than those with purple overall and 3200 recent.

 

And to add onto the topic, there are just as many deep purple overalls who only got there by playing the right tanks and stays there by only playing the right tanks.

Link to post
Share on other sites

I don't think it's possible for a green to get 3500 recent all of the sudden without having some major padding tanks (which is MaxL_1023 levels of play), which would more than certainly lift a teal up as well if it involves padding tanks. Like I said, you and Kewei are proof of that concept. Also sorry for being bad@reading. Plus, the changes in habits would be reflected over time if they can truly maintain that level of play.

 

And yes Med, I'm a shitter. That doesn't necessarily invalidate my point in making this topic though, does it?

It also doesn't stop me from shitpoasting breh. 4poGElb.png

Link to post
Share on other sites

I've thought about this every time I've been forced to use noobmeter and see the ~past 10k games wn8 or whatever it is. It would stop a large amount of recent padders, but at the same time diminish the morale of people with shit recent by making it more like raising two overalls. It doesn't matter how the metric is coded, or how the pretty colors are measured, people will still pad.

Link to post
Share on other sites

I've thought about this every time I've been forced to use noobmeter and see the ~past 10k games wn8 or whatever it is. It would stop a large amount of recent padders, but at the same time diminish the morale of people with shit recent by making it more like raising two overalls. It doesn't matter how the metric is coded, or how the pretty colors are measured, people will still pad.

That's more of a problem with WN8 than anything else though. At least WN9 is on the horizon by the looks of it, and it addresses several key issues with WN8. Plus, where's the drive to improve if everyone just drives E50s and pats each other on the back? Cmon guys, is that really any better than ARMED and IOC_1 huddling together to protest themselves not being shit?

Plus, a percent of overall battles is easier to raise than the whole enchilada. I don't know if I'll even be blue overall by the time I burn myself out of this shit game, but at least I'll know that reached the point where I can (hopefully) suppress my inner red to the point of playing like a purple.

Link to post
Share on other sites

It's accepted that a dark blue overall with light purple recents is likely better than a teal overall with deep purple recents, discounting all other factors, on the basis that they have more good gameplay habits (such as mechanical skills and quick judgement of the minimap) ingrained 

Except those good gameplay habits don't seem to be giving them the edge over their last 1000 games assuming none of the two are padding. 

Link to post
Share on other sites

That's more of a problem with WN8 than anything else though. At least WN9 is on the horizon by the looks of it, and it addresses several key issues with WN8. Plus, where's the drive to improve if everyone just drives E50s and pats each other on the back? Cmon guys, is that really any better than ARMED and IOC_1 huddling together to protest themselves not being shit?

People will know if you just pad E50 to have good WN8 same as if you were to play lights or pad lower tiers. People will know you're still a shit player and will call you out for being shit. 

Link to post
Share on other sites

i think the benchmarks for blue/uni/super uni would change as most peoples recents are higher than their overall, thus increasing the average skill level. so to maintain the 1%/.1%/.01% server population ratio the wn8 benchmark would increase. i would be curious of the actual values 

        Overall WN8      Recent WN8
1%          2208            2538   
.1%         2873            3470
.01%        3540            4488

Overall WN8: All EU accounts with 2k+ battles and 30-day activity, overall stats. The overalls are higher than usual, mostly because the WoTLabs percentiles are from the RU server where the average damage numbers are lower, tier for tier.

Recent WN8: 30-day, all EU accounts with at least 20 battles over the period.

 

Link to post
Share on other sites

Here's the problem I see with your concern.

In order to check recent WN8, you have to go to the Wotlabs page or see it in a siggy. Either way, you can get to the Wotlabs page.... then scroll down to the bottom, see if they are just padding or if they are actually good (by looking at how they do in ALL of their tanks individually), and have all of the answers you need to know.

WN8 can be skewed pretty easily. I know a barely purple player that has much better overall stats than me, but is only purple in TWO tanks, as opposed to me in a significant number of mine. His numbers are skewed because he has a SHITLOAD of battles in one tank, and the next most in the next. Anyone who actually cares to look that up can see if he's any good in his tanks, which quite honestly he's bad in anything that isn't those tanks.

But that's the thing. No one is going to recruit you off of just a siggy, or just the top part of your wotlabs page. This seems pointless to discuss IMO.

Link to post
Share on other sites
        Overall WN8      Recent WN8
1%          2208            2538   
.1%         2873            3470
.01%        3540            4488

Overall WN8: All EU accounts with 2k+ battles and 30-day activity, overall stats. The overalls are higher than usual, mostly because the WoTLabs percentiles are from the RU server where the average damage numbers are lower, tier for tier.

Recent WN8: 30-day, all EU accounts with at least 20 battles over the period.

 

where did you get these numbers from? thanks

Link to post
Share on other sites

where did you get these numbers from? thanks

WoT has a public API. It's fast enough to pull the entire server's worth of account/info and account/tanks data in under a day, which is sufficient for WN8 values. To get the recent values, you wait 30 days, pull the data again and subtract.

 

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...