Jump to content
TheLovePanda

Why Frags Should be Removed from the WNX Formula

Recommended Posts

Preface: I am not a mathematics contributor and have only a basic knowledge of how WNX works, so if this sounds completely ridiculous please feel free to neg and make me look stupid.

 

***

 

Frags should be removed from the WNX formula, leaving damage (dealt and eventually assisted) as the primary contributor to overall WNX score.

 

The reasons for this idea are not mathematical, as I am aware that frags have a high correlation with winning, but I think a convincing case may be made for this change.

 

I feel like the foundational question behind the WNX formula is "what factors contribute to a win?", assuming that the goal of the WNX metric is to rate how well a player contributes to the win of a match. Looking at the WN8 formula we can see that the two most prominent factors are damage and frags. But should they both be weighed so heavily? I say no.

 

Why is damage correlated with wins?

 

In my experience, most matches are won by elimination, that is, all tanks of one team are destroyed regardless of game type. I'm not sure that this is true across the board, but I suspect it is (even if it is not, there are still valid points to be made). If it is true, then the question must be asked, "what stands between your team and a win over the enemy team?" Some may say kills, but this is not actually true. Though all tanks must be fragged to win by elimination, the only way to get a secure a frag is to move the enemy HP to 0. If all enemy tanks were down to 1 hitpoint, a low tier scout could theoretically load HE and kill them all quite easily. So, in reality, it is clear to see that the HP pool of the enemy team is a far more significant barrier than just the individual frags.

 

From this perspective the correlation between damage and winning is very logical, seeing that as more HP is eliminated from the enemy team the more likely your team is to get kills and to win.

 

Why are frags correlated with wins?

 

Assuming that most matches are won by elimination, frags are made a prerequisite to winning, and will therefore obviously be correlated with wins. It would be quite difficult to get 15 kills and lose.

 

However, kills are only possible by damaging the enemy, and therefore are a byproduct of damage and not an equal factor. In this sense it seems obvious that both would be highly correlated with winning, but one is a precursor to the other, and in fact supersedes the other in terms of practical significance. Above I said that as your team gets more damage, frags become more likely, and therefore a win becomes more likely, but the base factor in this chain is damage, and other factors depend upon it for significance.

 

TL;DR

Frags are then, in my view, incidental, and do not by themselves contribute hardly anything to a win. The only situation in which frags contribute highly to a win is when a player eliminates a priority target which threatens the win rather than choosing to farm damage. However, this situation would be impossible to factor in mathematically, and unless some arbitrary limiting factor were to be incorporated to lessen the impact of frags on the formula the only logical conclusion is to remove it completely, leaving damage as the primary factor of the WNX metric.

Share this post


Link to post
Share on other sites

If this were the case I think so many people would pick up the bad habit of firing at a full health target rather than the 1 shot target next to it- just to farm wnx. I'd much rather see people removing guns from the field. 

Share this post


Link to post
Share on other sites

IIRC frags are a small part of the WN8 calc.

 

It was terrible WN7, where the best way to pad the number was to snipe kills (or farm kills in tier 1) and cheese the average tier while damage didn't matter as much. One of the goals of WN8 was to fix that problem and Praetor77 did extensive research to come up with the formula.

Share this post


Link to post
Share on other sites

So wn8 whore pubbies can even more whore wn8 in a negative way and spam at full health tanks in their T 62A while there is an enemy T57 on 50 health 5sec from clipping my bat chat? No. No thank you.

 

This is about winning, not whether or not you get clipped out in your bat chat.

IIRC frags are a small part of the WN8 calc.

 

It was terrible WN7, where the best way to pad the number was to snipe kills (or farm kills in tier 1) and cheese the average tier while damage didn't matter as much. One of the goals of WN8 was to fix that problem and Praetor77 did extensive research to come up with the formula.

 

WN8 = 980*rDAMAGEc + 210*rDAMAGEc*rFRAGc + 155*rFRAGc*rSPOTc + 75*rDEFc*rFRAGc + 145*MIN(1.8,rWINc)

 

As you can see rFRAGc occurs as a contributing factor in 3 of 5 total calculations.

Share this post


Link to post
Share on other sites

If kills somehow incidental "and do not by themselves contribute hardly anything to a win."

 

The damage is incidental as it just so happens that RNG allows my shot to hit and does not contribute hardly anything to a win. Capping should be 100% of wn8.

Human creation is incidental and hardly adds to a win, wn8 shouldn't exist because it was just chance that humans got here.

NEGREP BY ANAL WARLORDS PLIS.

Share this post


Link to post
Share on other sites

If kills somehow incidental "and do not by themselves contribute hardly anything to a win."

 

The damage is incidental as it just so happens that RNG allows my shot to hit and does not contribute hardly anything to a win. Capping should be 100% of wn8.

Human creation is incidental and hardly adds to a win, wn8 shouldn't exist because it was just chance that humans got here.

NEGREP BY ANAL WARLORDS PLIS.

 

Any shitter can put the last shot in. Not any shitter can get 4000+ damage consistently in tier 10.

Share this post


Link to post
Share on other sites

Clearly nobody can be asked to actually read what I wrote or make a single bit of serious discussion. 

I dunno man sounds like you're the one who needs to try reading.

 

This is about winning, not whether or not you get clipped out in your bat chat.

and clearly if his batchat gets clipped out because you didnt kill the fucking t57 then you are now more likely to lose!

 

And winning has a fairly strong correlation with rFRAGc

 

/thread.

Share this post


Link to post
Share on other sites

WN8 = 980*rDAMAGEc + 210*rDAMAGEc*rFRAGc + 155*rFRAGc*rSPOTc + 75*rDEFc*rFRAGc + 145*MIN(1.8,rWINc)

 

As you can see rFRAGc occurs as a contributing factor in 3 of 5 total calculations.

 

But doesn't mean it's responsible for 60% of your WN. If i remember right from when it was being worked on, frags are weighted against other values (and possibly capped?) so you can't get good WN just from stealing kills, but can't be removed from the formula because it does correlate to winning. Damage is already the primary contributor to WN8.

 

Someone that knows the formula better should be able to explain better how it works, if it's not already explained in the WN wiki.

Share this post


Link to post
Share on other sites

But doesn't mean it's responsible for 60% of your WN. If i remember right from when it was being worked on, frags are weighted with other values (and possibly capped?) so you can't get good WN just from stealing kills, but can't be removed from the formula because it does correlate to winning. Damage is already the primary contributor to WN8.

 

Someone that knows the formula better should be able to explain better how it works, if it's not already explained in the WN wiki.

 

Right, but if you get 0 kills then those 3 factors are completely eliminated from the formula which has a HUGE impact on WN8. A 4000ish damage game in some tier 10s with 0 kills is only worth 2000ish WN8, even if the expected damage is only around 2000. I would call that an exceptionally good match, and the fact that it only gets a blue WN8 rating seems off.

Share this post


Link to post
Share on other sites

Right, but if you get 0 kills then those 3 factors are completely eliminated from the formula which has a HUGE impact on WN8. A 4000ish damage game in some tier 10s with 0 kills is only worth 2000ish WN8, even if the expected damage is only around 2000. I would call that an exceptionally good match, and the fact that it only gets a blue WN8 rating seems off.

WN8 should not be looked at over a single game. You are using a tool wrong like trying to unscrew a screw with a mallet.

If you are consistently (read averaging) doing 4k damage and no kills you are one heck of an outlier.

Share this post


Link to post
Share on other sites

Panda, I cannot into math but I can into PvP and here's a protip: A 50 HP tank does as much damage as a 2000 HP tank but a 0 HP tank does 0 damage 'cause it's dead. What you're essentially saying is that shooting 15 tanks until all of them are at 50 HP has the same value as those 15 tanks focus-firing and killing your dudes off one by one. I know which 15 players are contributing more to winning the battle and it's not the team where everyone has nearly 2k dmg done with 0 kills.

 

It may be possible that the kill/damage ratio needs some tweaks but that requires math skills, data and more insight than I have.

Share this post


Link to post
Share on other sites

Right, but if you get 0 kills then those 3 factors are completely eliminated from the formula which has a HUGE impact on WN8. A 4000ish damage game in some tier 10s with 0 kills is only worth 2000ish WN8, even if the expected damage is only around 2000. I would call that an exceptionally good match, and the fact that it only gets a blue WN8 rating seems off.

 

4000 damage with 0 kills seems to be an outlier and not something common enough that it should be worried about in the formula. Remember that WN8 was not made to be used in a per game basis and it doesn't work well in either extremes of player skill.

Share this post


Link to post
Share on other sites

I've run a lot of models trying to evolve and improve WN8 into WN9. Every model that works well in correlating with winning has frags in it somewhere. More recent models even have frags weighted higher than damage, so that might feature in WN9.

 

So, what I can state with absolute confidence, based on hours of running various models in R, is that frags are a key component of WN, so no, they are not likely to come out of WN anytime soon.

Share this post


Link to post
Share on other sites

More recent models even have frags weighted higher than damage, so that might feature in WN9.

 

That could be bad if weighted poorly. Damage farmers might piss some people off, but it's better than people holding their fire to wait for a kill shot.

Share this post


Link to post
Share on other sites

Clearly nobody can be asked to actually read what I wrote or make a single bit of serious discussion. 

You didn't do your homework on WN8. The discussion happened over a year ago, together with the statistical justification for incorporating frogs.

 

Choo choo!

neg-train-o.gif

Share this post


Link to post
Share on other sites

I've run a lot of models trying to evolve and improve WN8 into WN9. Every model that works well in correlating with winning has frags in it somewhere. More recent models even have frags weighted higher than damage, so that might feature in WN9.

 

So, what I can state with absolute confidence, based on hours of running various models in R, is that frags are a key component of WN, so no, they are not likely to come out of WN anytime soon.

 

Wouldn't it be a problem if damage and frags weren't independent factors? I feel like if you did a multiple regression analysis of wins versus all of the WN8 ratios your residuals would not be random.

 

nvm

Share this post


Link to post
Share on other sites

That could be bad if weighted poorly. Damage farmers might piss some people off, but it's better than people holding their fire to wait for a kill shot.

 It'd be offset by refinements in dmg per tank (now avail in the API), likely with separate formulas per class.  

 

Haters or not, each iteration of Eff -> WN4/6 -> WN7 -> WN8 has proven harder for the less skilled to manipulate. Remember the legions of capfasters? Remember the Hellcat brigade? Remember the Tier 1/Tier 10 heroes? Autism is now forced to basically only play his tier 5 and 6 scouts to pad rating now, and even in doing so he can no longer hang with the actually good players who can pad WN8 near the limits, or with actually good players who play those same tanks (paging B0B_Ross). I'm not saying the metric is perfect, just pointing out that each time its been revised less and less people are able to manipulate it. 

 

And like spencer said...you missed this train OP. Check the wiki, comb through the development thread, but frags are in because the data says so. 

Share this post


Link to post
Share on other sites

This is about winning, not whether or not you get clipped out in your bat chat.

 

WN8 = 980*rDAMAGEc + 210*rDAMAGEc*rFRAGc + 155*rFRAGc*rSPOTc + 75*rDEFc*rFRAGc + 145*MIN(1.8,rWINc)

 

As you can see rFRAGc occurs as a contributing factor in 3 of 5 total calculations.

 

Yes, but rFRAGc doesn't occur as a first-order term; only damage does.

That means that the contribution of kills to WN8 increases as spotting and damage increase.

 

I'm quite familiar with the logistic regression methods used in predictive modeling such as WN8...

And this kind of relation seems quite reasonable to me.

Share this post


Link to post
Share on other sites

You didn't do your homework on WN8. The discussion happened over a year ago, together with the statistical justification for incorporating frogs.

 

Choo choo!

 

 

Do you remember what thread it was in?

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...