Jump to content
bjshnog

⟪WN8⟫ Development / Resources

Recommended Posts

Just inputting here. 

 

This is probably too complex, but since noobmeter's interval feature (to a certain extent), offers a look on the number of battles played in certain tanks between certain periods of time, I think it would be possible to use that feature to root out people padding current FOTM or OP tanks by attaching rating coefficients to specific periods of time. 

 

So for example, someone with 300 games, or a vast majority of their games (250/300) in 8.6+ in a TD (tanks that are overpowered right now) like the Foch 155 or Object 268, should get a lower rating coefficient than someone with 300 games, or a vast majority of their games (250/300) before 8.6. 

 

Or we could be a bit more "precise", and just attach rating coefficients to very specific intervals (lots of info-gathering to do, mainly dates of patch releases that made tanks OP, and dates of patch releases that nerfed tanks).

 

So, any stats attached to games played before the release of 8.6 for the Object 268 or Foch 155 should have a lowered rating coefficient, and any stats attached to games played after their nerfs, or the introduction of a new mechanism that nerfs them indirectly, should have an heightened or restored/average rating coefficient. 

 

As for tanks that became OP and were nerfed while the interval feature was working (M4 Derp), simply lower the rating coefficient for the games played during the time-frame while it was OP, and raise the rating coefficient for the games played before and after the nerfs. 

 

As for tanks that will become OP later, same idea, lower the rating coefficient while it's OP, but raise it once it gets nerfed. 

 

Etc. 

 

The main problem though, is that you can't root out tanks that were OP before noobmeter started tracking intervals, like the early M48A1 or T110E5 (Russians had 4500~ DPG in them before their first round of nerfs, and not even 4000~ in any of the other tier 10s at the time (not even TDs) to give you an idea of how OP they were).

 

Also, the rating coefficients would have to be determined by subjective means, i.e. purple poaster council.

 

AND also, I'm not sure how you would code it. 

Link to post
Share on other sites

Just inputting here. 

 

This is probably too complex, but since noobmeter's interval feature (to a certain extent), offers a look on the number of battles played in certain tanks between certain periods of time, I think it would be possible to use that feature to root out people padding current FOTM or OP tanks. 

 

So for example, someone with 300 games, or a vast majority of their games (250/300) in 8.6+ in a TD (tanks that are overpowered right now) like the Foch 155 or Object 268, should get a lower rating coefficient than someone with 300 games, or a vast majority of their games (250/300) before 8.6.

 

As for tanks that became OP and were nerfed while the interval feature was working (M4 Derp), simply lower the rating coefficient for the games played during the time-frame while it was OP, and raise the rating coefficient for the games played before and after the nerfs. 

 

As for tanks that will become OP later, same idea, lower the rating coefficient while it's OP, but raise it once it gets nerfed. 

 

Etc. 

 

The main problem though, is that you can't root out tanks that were OP before noobmeter started tracking intervals, like the early M48A1 or T110E5 (Russians had 4500~ DPG in them before their first round of nerfs, and not even 4000~ in any of the other tier 10s at the time (not even TDs) to give you an idea of how OP they were). 

 

Also, I'm not sure how you would code it. 

 

There would no way to measure it or put it in the rating, because we don't have historical stats over the database sample for each tank and we don't have historical stats for each player's stats in those tanks. Even on Noobmeter, a player might have played a tank a lot when it was OP, but until they start tracking it on Noobmeter, the times they played that tank will all coincide at the point at which they began tracking on Noobmeter and the data won't be separable.

Link to post
Share on other sites

Which is what the last paragraph said. 

 

But you can still use it to root out tanks that became OP during the intervals, or tanks that are currently OP. 

 

Still better than nothing. 

 

It would have no application apart from maybe a note saying "this tank was once OP or has become OP at some point, but there is no way to tell when this player played it, so we can't adjust for it, sorry."

Link to post
Share on other sites

It would have no application apart from maybe a note saying "this tank was once OP or has become OP at some point, but there is no way to tell when this player played it, so we can't adjust for it, sorry."

 

And we're back to the noobmeter intervals, which we can use to root out tanks that are currently OP or were OP during the intervals, and attach lower rating coefficients to any stats gained during that period

 

(i.e. 3000 DPG in a HEAT-54 after it got its HEAT rounds and before the HEAT round nerfs would be lowered to 2700 DPG when used to calculate WN rating by attaching a rating coefficient of 0.9 or something). 

 

Correct me if I'm wrong, but when I looked at the noobmeter intervals, I saw average DPG during an interval laid out, so I'm assuming noobmeter also tracked individual tank DPG/s, and that it's also possible to extract that individual tank DPG data during these intervals.  

 

I also saw that individual tank WR was tracked during that period as well. 

 

Now, if we can't track individual DPG, then I'll concede, but if we can. Well, the coding's probably going to be monstrously complex, but it can be done, can't it?

Link to post
Share on other sites

And we're back to the noobmeter intervals, which we can use to root out tanks that are currently OP or were OP during the intervals, and attach lower rating coefficients to any stats gained during that period

 

(i.e. 3000 DPG in a HEAT-54 after it got its HEAT rounds and before the HEAT round nerfs would be lowered to 2500 DPG when used to calculate WN rating). 

 

Correct me if I'm wrong, but when I looked at the noobmeter intervals, I saw average DPG during an interval laid out, so I'm assuming noobmeter also tracked individual tank DPGs, and that it's also possible to extract that individual tank DPG data during these intervals.  

 

We won't only be using noobmeter. The only stats we can use are those supplied by WG through their API.

 

Besides, we don't have historical data for each tank anyway, so we can't do it. Noobmeter doesn't track stats in individual tanks.

Link to post
Share on other sites

Well fuck, would be great if someone started collecting data to preserve historical stats between patches/fixes, we won't be able to root out past padders, but at least we can root out future padders.

 

It wouldn't be that tough either would it? There are tons of sites like wot-news or wotlabs that literally update tens of thousands of tanker stats by the minute/hour/week, and seeing as patches usually have months in between them and seeing as we would only be looking at historical stats at the moment before the patches...

Link to post
Share on other sites

Its not easy. That results in HUGE, and I mean HUGE databases... noobmeter is up to several terabytes already... also, since this rating is meant for use in xvm, they would have to be the ones who implement such solution. Its beyond our scope with WN8 anyways.

Link to post
Share on other sites

I'm pretty sure the T-50-2 is now counted as the MT-25, so those stats should be merged.

Link to post
Share on other sites

Maybe you could make something up with your current formula for people to check their wn8 

 

Then purple players can check a bunch of peoples wn8 and let you know pretty quick that this and that players score is bs.

 

 

Really all the formulating in the world is worthless if good players are rated badly and stat manipulators highly.

 

 

 

This new Wn8 imo looks like it will only reward playing tanks that are bad for the majority of the population.  I think that damage by tier was a fine judge of skill.  If your doing over 2500 damage in 8's 3,000 in 9's and 3500 in 10's and don't have a high wn8 there's a problem there, no matter what tank you drive.

 

From what I gather in wn8 there will be very few unicums unless they did well on a20's..etc crap tanks.

Link to post
Share on other sites

Maybe you could make something up with your current formula for people to check their wn8 

 

Then purple players can check a bunch of peoples wn8 and let you know pretty quick that this and that players score is bs.

 

 

Really all the formulating in the world is worthless if good players are rated badly and stat manipulators highly.

 

 

 

This new Wn8 imo looks like it will only reward playing tanks that are bad for the majority of the population.  I think that damage by tier was a fine judge of skill.  If your doing over 2500 damage in 8's 3,000 in 9's and 3500 in 10's and don't have a high wn8 there's a problem there, no matter what tank you drive.

 

From what I gather in wn8 there will be very few unicums unless they did well on a20's..etc crap tanks.

 

If the A-20 is so bad, then unicums will also not get stellar scores in it. Unicums are going to be redefined in terms of WN8 score anyway. Also, if you use OP tanks (Foch 155), then you're not actually playing better, you're just playing an OP tank. Sure, you might win more, but once again, you're just playing an OP tank.

Link to post
Share on other sites

So basically you are trying to encourage people to play shitty tanks more...got it.

 

 

I don't know many good players that say man I love me some M48 Patton or IS7.   Are they bad tanks?  No not really they just aren't as fun to play so most people don't.

 

But if I understand how WN8 works your going to need 4k dpg and 2.25 kills in a T57 heavy to not get penalized for playing an "OP" tank.

 

Theres a few people on NA that can do that...so I guess the new trend will be playing boring tanks.

 

 

You never hear about how OP the 268 is or the E4 and E-100 (but they are)..apparently they just don't stick in your mind when you die to them.

Link to post
Share on other sites

So basically you are trying to encourage people to play shitty tanks more...got it.

 

 

I don't know many good players that say man I love me some M48 Patton or IS7.   Are they bad tanks?  No not really they just aren't as fun to play so most people don't.

 

But if I understand how WN8 works your going to need 4k dpg and 2.25 kills in a T57 heavy to not get penalized for playing an "OP" tank.

 

Theres a few people on NA that can do that...so I guess the new trend will be playing boring tanks.

 

 

You never hear about how OP the 268 is or the E4 and E-100 (but they are)..apparently they just don't stick in your mind when you die to them.

 

No.

Link to post
Share on other sites

So basically you are trying to encourage people to play shitty tanks more...got it.

 

 

I don't know many good players that say man I love me some M48 Patton or IS7.   Are they bad tanks?  No not really they just aren't as fun to play so most people don't.

 

But if I understand how WN8 works your going to need 4k dpg and 2.25 kills in a T57 heavy to not get penalized for playing an "OP" tank.

 

Theres a few people on NA that can do that...so I guess the new trend will be playing boring tanks.

 

 

You never hear about how OP the 268 is or the E4 and E-100 (but they are)..apparently they just don't stick in your mind when you die to them.

 

The idea is not to encourage anything. The idea is to be able to fairly compare a player who played 1000 games on A-20 and another who played 1000 games on M8A1.

 

Tier based damage calculation was not good enough. Would you consider equally skilled, a player who only played the A-20, with 1000 avg damage and another who only played Matilda, with 1000 average damage? Same for IS-7 and Foch155... tier based is just not precise enough.

 

The idea is to be able to compare player skill regardless of tank selection.It seems you are really not understanding how WN8 works...

 

 

It´s weird to see you be non-constructive and criticize without ANY information to back you up... at all. Also, regarding your asking for different player´s rating to be able to give WN8 the eye-test, I am on it, I just havent been home the entire weekend...

Link to post
Share on other sites

I get the concept but all it will do is make people broaden their tank selection.  Your going to have people playing tanks like the superpershing to farm wn8.  The expectations from that tank are low but it's actually easy to own in.

 

Oh well my 2 cents.

 

 

But people will always game the game.  

Link to post
Share on other sites

I read a bit of this thread and my head is spinning at the absurdity I saw typed; here are my 2 cents.. if you want to devalue WN as a rating platform, do go ahead and start screwing with it based on the tanks people play.  We laughed at how irrelevant WG's rating is, yet here we you are going down the same path.

 

It is beyond arrogant to try to dictate what tanks people play in order to get ratings.. for one it will backfire, and for second, it will further devalue platoons, which are already seen as stat padding means on their own, regardless of tanks.

 

The power of WN comes from people considering it an accurate and FAIR measure ones APPROXIMATE skills. When you begin to ADJUST IT according to the NEEDS of CERTAIN players (which is what wn8 is really, lets be frank) it will hold as much weight as well.... paper.

 

feel free to neg :)

Link to post
Share on other sites

I read a bit of this thread and my head is spinning at the absurdity I saw typed; here are my 2 cents.. if you want to devalue WN as a rating platform, do go ahead and start screwing with it based on the tanks people play.  We laughed at how irrelevant WG's rating is, yet here we you are going down the same path.

 

It is beyond arrogant to try to dictate what tanks people play in order to get ratings.. for one it will backfire, and for second, it will further devalue platoons, which are already seen as stat padding means on their own, regardless of tanks.

 

The power of WN comes from people considering it an accurate and FAIR measure ones APPROXIMATE skills. When you begin to ADJUST IT according to the NEEDS of CERTAIN players (which is what wn8 is really, lets be frank) it will hold as much weight as well.... paper.

 

feel free to neg :)

 

And this is why altering the data of certain tanks is a problem. People will start to think that you are catering to certain players. I think you should just go with the data you extracted from vbaddict without trying to alter it.

And to the t57-heavy point. You have to do about 3225 avg. damage to get a 2400 wn8 rating? and 2400 should be similar to a 2100 wn7 score? I am doing about 3050 avg. damage in my t57 and i don't consider myself a true 2100 wn7 player. So i guess this should be about right?

Link to post
Share on other sites

I read a bit of this thread and my head is spinning at the absurdity I saw typed; here are my 2 cents.. if you want to devalue WN as a rating platform, do go ahead and start screwing with it based on the tanks people play.  We laughed at how irrelevant WG's rating is, yet here we you are going down the same path.

 

It is beyond arrogant to try to dictate what tanks people play in order to get ratings.. for one it will backfire, and for second, it will further devalue platoons, which are already seen as stat padding means on their own, regardless of tanks.

 

The power of WN comes from people considering it an accurate and FAIR measure ones APPROXIMATE skills. When you begin to ADJUST IT according to the NEEDS of CERTAIN players (which is what wn8 is really, lets be frank) it will hold as much weight as well.... paper.

 

feel free to neg :)

 

Either you read wrong, or you are trolling... or you're an idiot. When WN8 is finally released and the description is released, feel free to come back.

 

I really should stop throwing insults around, but it really irritates me when someone says WN8 is going to be BS and/or as irrelevant as the new WG rating without actually knowing anything about it.

Link to post
Share on other sites

Buncha bullshit at the end of this thread. Either try to read, learn to read, or don't fucking post. We don't have to tolerate it here, thankfully. (I prefer this type of verbal warning before I delete your future drivel, it helps make it clear I'm not censoring dissent, but idiots)

 

lifeless, bjshnog's assessment stands. Are you clueless or trolling?

 

Belial - if you think people are going start "farming" WN8 on bad tanks...well you're paranoid as fuck. Sure, if I were to jump into a Super Pershing I'd be able to exceed the standards set, and it might bump my WN8. But I personally would have to be a sick fuck to play a shitty no-carry SP to 'inflate' my WN8*. Is it really inflation since I am already good at WoT? Secondly, which is is easier - exceeding the Super Pershing standards by a good margin, or playing tier 10 and tier 1 in a 2 to 1 ratio and pushing out 2200+ WN7 scores with only marginal performances at both tiers. Praetor made it very very clear why average damage per tier was insufficient.

 

*And sure, someone might do it. But we're going to keep re-norming** the performance standards per tank, so any attempts to manipulate will be disarmed...

 

Link to post
Share on other sites

And this is why altering the data of certain tanks is a problem. People will start to think that you are catering to certain players. I think you should just go with the data you extracted from vbaddict without trying to alter it.

And to the t57-heavy point. You have to do about 3225 avg. damage to get a 2400 wn8 rating? and 2400 should be similar to a 2100 wn7 score? I am doing about 3050 avg. damage in my t57 and i don't consider myself a true 2100 wn7 player. So i guess this should be about right?

 

Once again, I am not catering to anything or anyone. If we had 1000000 dossier files to analyze, we wouldn´t need to pry and analyze stats for each tank so closely to determine a fair balance amongst different tanks. Sadly, we do not, and we have to make the most of the data we have. I am not "altering" ANY data. Whatsoever. All the data is being extracted from the vbaddict dossier DB. 

 

The thing is, as I have explained several times now, I used a metric (tankWN8/playerWN8) to determine if we are measuring stats for players of the same level when comparing two tanks (or rather 315 tanks). This is good enough for 80% of tanks, but when you analyze the data using other metrics, some tanks clearly stand out, as having uncomparable stats. This is due to a million different factors which I have already gone over, nerfs/buffs, matchmaker changes, etc.

 

The road I have taken here is to go ahead and let the wotlabs community (specially purples, since their understanding of game mechanics and balance is superior to those of the average player) take a look at these stats and point out statistical outliers. Once they do, I take a much closer look at the stats for that tank, and the players in vba who have played them (although, all the player data is anonymous, I only have a playerID, which has nothing to do with WG player ID).

 

 

 

Per-tank balance is necessary for accurate protrayal of player skill. Like I said, if the same player plays 10000 games on the Hellcat, opens a new account, plays 10000 games on the ARLv39, on another account plays 10000 games on M12, or on AMX12t, or whatever. His WN7 would be MILES apart on each case, although his avg tier would be exactly the same.

 

 

Belial, where we agree is that padders might switch to playing A-20 instead of M8A1. I really don´t think it will happen. Two reasons: 1- Playing bad tanks is not fun. 2- They will realize they can´t really best the "expected" stats for A-20 too easily. 3- We will keep an eye on vbaddict ratings per-tank. That is, we will monitor tankWN8/playerWN8 (which is also what we use to derive expected stats) for all tanks to make sure our expected stats were not wrong. If our stats for A-20 were too low, and unicums can  and do abuse it, we just recalculate the expected stats, and the problem should be solved.

 

 

WN8´s intention is in no way to dictate what tanks players should play, maybe even the exact opposite. You could propose OP tanks in their tier are overplayed due to 2 reasons. 1- It´s fun to own in games, and do very well. 2- Farming WN7.

You could also propose some tanks are underplayed due to WN7 rating taking a plunge when playing them too much (light tanks). WN8 will make it fair, so that your WN8 only changes with how well you play your tanks, and not WHAT tanks you choose to play.

Link to post
Share on other sites

The power of WN comes from people considering it an accurate and FAIR measure ones APPROXIMATE skills. When you begin to ADJUST IT according to the NEEDS of CERTAIN players (which is what wn8 is really, lets be frank) it will hold as much weight as well.... paper.

 

I assure you WN8 will be more accurate and more fair than WN7. I am not adjusting anything according to anyone´s needs... this is a blatant lie. Please show data to back up your claim. I can show you pages and pages of data to back up mine.

 

 

If you think community development is "ADJUSTing IT according to the NEEDS of CERTAIN players", then that is just plain dumb. The entire development of the metric is being clearly posted here, in an absolutely transparent way. Would you prefer me to simply do whatever the heck I think is best, without consulting, contrasting and discussing the data and results?  :thumbdown: 

Link to post
Share on other sites

bjshnog, it seems you really care about the difference between setting the baseline at zero countable contribution (no damage spots kills etc) vs zero effective contribution (no amount of damage spots kills sufficient to influence the match beyond a bot). 

 

But I have to ask...do you reject the idea that there are thresholds below which contributions are simply noise? (per the IS case from page 12 or 13)

 

Ultimately, it comes down to whether we're trying to measure "activity" vs "ability to contribute enough to influence battle outcome". Praetor and I favor the latter. The former is already covered in the basics of the Service Record. You want to award points for activity, but I have a hard time seeing why. I am open to arguments, and I want to make sure we go through the idea thoroughly. 

 

Also your thread on the WGNA forums was very...bjshnog vs Praetor, which seems to be an unfair characterization of the debate. It should be based on the academics of the issue, not the personal side of things...

Link to post
Share on other sites

Incorporating a per-tank-average rating would also go a ways towards solving the light tank rating problem. 

Link to post
Share on other sites

Incorporating a per-tank-average rating would also go a ways towards solving the light tank rating problem. 

 

 

That basically the crux of WN8, the real big difference.

 

As you are an experienced and able tanker, please click back a few pages and check out the per-tank-averages that Praetor has extracted from the vbaddict databse, and look over some sections? The vbaddict db isn't huge, and so we're looking for visual/manual validation from a crowd-sourced panel of good tankers (and you've just gotten yourself nominated!)

Link to post
Share on other sites

So if i am understanding this correctly, I should go back and rebuy tanks that i played in my noobish days to help my current WN8 rating more accurately reflect my current skills?  Or will it not really matter?

 

Does say for example my M3 Lee stats from like my first 400 games ~= my last 400 games in my T49 or T54 as far as my overall skill rating calculation? Do they carry the same weight?

 

Don't get me wrong, simply a question, as i have been reading through this from time to time to get a further understanding.

 

 

Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...