Jump to content
bjshnog

⟪WN8⟫ Development / Resources

Recommended Posts

Fair enough on the time required. I think the community as a whole is very appreciative of the effort you guys put in to continually refine and improve these metrics.

Any thoughts then in having a 'post 8.8' refined recent category?

 

IMHO, it´s not worth the huge amount of work that would entail. I just don´t have the time right now. I´m struggling as it is to finish WN9 (two value system) within a reasonable time constraint. Also, the samples would be tiny, and probably not large enough to build a solid model.

Link to post
Share on other sites

Bigger than the existing recent tho.

But yes, if the time isn't there I guess we'll need to make do.

Link to post
Share on other sites

Thats because it is the most important factor in winning games as preator explains here:

 

His and my own analysis was not rigorous with regards to data cleaning. You can't take 500k rows (not battles) at random from a dataset with 14million that is for all technical purposes stratified/clustered by battlearenaid and userid.

 

I really didn´t do an exhaustive analysis, because I really have better things to do with my time, I think Xelos is doing an in-depth analysis of the spotted damage data.

 

Since historical spotted damage will never be available, I saw no sense in going into a detailed analysis for something that is never going to happen. I merely wanted to show that we are not missing out on that much due to not having spotted damage available (specially not for account-wide WN8 measurement). Also, as I said above, it´s a whole different picture for single-battle outcome prediction as opposed to account-wide measurement. 73% prediction power in single games for DMG alone translates to about +-2.75% WN8 winrate error, while adding kills, def and spots reduces that to +-1.84%. One could only speculate what spotting damage would reduce that to. Extrapolating, my guesstimate would be about +-1.4% to +-1.7%.

 

Please just hold off on using any of this, my own very prelim analysis was violating assumptions using the entire dataset albeit less than using 500k random rows.

Link to post
Share on other sites

No, you are not assigned any cap points if you lose the battle.  Check the battle results after the game.

 

There was a point in time where cap points only counted if the win was by cap, but that is no longer the case.

 

I confirmed but forgot to post it.  If you lose in a cap race, the losing team/players do not get any cap points.  I think the cap points that you saw in your analysis was due to kills after cap was completed.  In a small portion of the games the last tank is killed after cap is finished, and in that case the capping team loses to the killing team.

Link to post
Share on other sites

I confirmed but forgot to post it.  If you lose in a cap race, the losing team/players do not get any cap points.  I think the cap points that you saw in your analysis was due to kills after cap was completed.  In a small portion of the games the last tank is killed after cap is finished, and in that case the capping team loses to the killing team.

 

Okay i understand what your saying.

Link to post
Share on other sites

Can you guys please increase the Waffenträger auf E 100's expected damage by 1,500

Thanks

 

It wouldn't be an expected damage estimate... it would be expected damage +1500.

Link to post
Share on other sites

Exactly what waffle players deserve.

 

The problem is everyone keeps asking for account-wide analysis instead of per tank(or tier) metric.

Link to post
Share on other sites

can someone please explain to me why every single source for WN8 ratings is different.

I track my per game on VAddict 

I look at my XVM numbers and my Wot labs and Noobmeter numbers.

They are all slightly different. 

My problem is that I have been in a slump and I want to track my daily progress so that I can improve my game.

Link to post
Share on other sites

The session summary in vbaddict flattens out the winrate component, it uses the expected winrate of your tanks rather than your actual winrate for the session.

 

The recent performance on wotlabs and noobmeter use a slightly different definition of recent, so they're including a different range of games.

 

Though even then there are some differences.  I think wotlabs gives me about 6-7 more WN8 than noobmeter for my overall.

Link to post
Share on other sites

I haven't read the entire thread, so maybe this is something you have already considered.  Has there been any thought given to rewarding potential damage received?  Maybe it's not possible like with spotting/assisted damage?  My guess is that adding some kind of factor for potential damage received in a game may help address the concern that WNX promotes something of an undesirable play style (i.e. hang way back and snipe).  Rebalance the rewards so that tanks who do the leg work for damage farmers receive more credit and the hide/snipe only types get less reward.  Could this be done?  Would it be worthwhile?  Please take this as constructive criticism, but I'd compare the current system to analyzing a baseball player for offensive performance only while ignoring defensive contribution.  And after all, runs are runs whether you score them while hitting or prevent them in the field. 

Link to post
Share on other sites

Sure the value of a run scored is marginally greater than a run prevented since it takes 1 to win, but ignoring run prevention is overlooking (to pull a number out of my ass) about 49.5% of the game.  To relate it back to internet tanks: when too many people play the maximize wnx game and hang back to snipe, the red team closes the ground and the magical hang at the edge of spotting/draw range ploy no longer works.  WNX only rewards the vital (to continue the analogy) defensive aspect of the game indirectly. 

Link to post
Share on other sites

Sadly the WN devs have indicated they are a bit short on time so unlikely that WN will ever incorporate things like spotting damage or pot dmg received (if that's even available to them). I'm hoping someone with the time puts their hand up to help round out the metric as more of an influence metric rather than a damage one.

Link to post
Share on other sites

Sadly the WN devs have indicated they are a bit short on time so unlikely that WN will ever incorporate things like spotting damage or pot dmg received (if that's even available to them). I'm hoping someone with the time puts their hand up to help round out the metric as more of an influence metric rather than a damage one.

 

Spotting damage is not really a matter of time issue, but issue with WG API.

Link to post
Share on other sites

Spotting damage is not really a matter of time issue, but issue with WG API.

The good ol' API. I realise it's won't be a career thing, but my understand is that spotting damage is available in the API post 8.8. I could be wrong on that so someone more familiar with the APIs output would need to confirm. But by using it, a more accurate recent 'WN8+' could be created. If the API doesn't supply it post 8.8 then yeah, it's a moot point at this stage.

Link to post
Share on other sites

Spotting damage is not really a matter of time issue, but issue with WG API.

 

Furthermore, WG has not recorded spotting damage from before 8.8, so it's incomplete data anyway.

Link to post
Share on other sites

Furthermore, WG has not recorded spotting damage from before 8.8, so it's incomplete data anyway.

 

The game has changed so radically from version to version I'd almost be interested in seeing a stat reset at version 1.0.

Link to post
Share on other sites

The game has changed so radically from version to version I'd almost be interested in seeing a stat reset at version 1.0.

 

But... that's even more incomplete...

Link to post
Share on other sites

But... that's even more incomplete...

 

Sometimes it is necessary to start over with new data when one can no longer control for numerous changes inherent in the data set.  =(

Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...