Jump to content
WarshipsToday

Warships Today Rating - a WoWS rating, and warships.today

Recommended Posts

3 minutes ago, Tedster59 said:

yeah, that looks nice.  my only real want is for there to be a way to give the website my private link so that I could keep my profile as "link only" to block any XVM type mod, but have the site track my stats and update the sig.

I don't think there has been a working in-game stats mod for months.  I suppose people can still look you up manually during the load-in.

Share this post


Link to post
Share on other sites
1 hour ago, Mesrith said:

I don't think there has been a working in-game stats mod for months.  I suppose people can still look you up manually during the load-in.

There's this, though it requires extra effort to set up. But I just assume everyone private on the enemy team is good and everyone private on my team counts as already sunk (just like most NC players).

Share this post


Link to post
Share on other sites

Wow my signature makes me look better than I really am (last night barring because I could hit the inside of a barn from the inside and I had one game that crashed and couldn't get back in till near the end of the battle)

Share this post


Link to post
Share on other sites
7 hours ago, WarshipsToday said:

Mnemon, yours should be OK in a few hours. It is still cached, but a fix was pushed out that fixes yours.

Yeah works now. I wonder what the issue was.

Share this post


Link to post
Share on other sites

Something about the WTR calculations at tier 10 doesn't feel right to me.  I think that everyone would agree that achieving a 2000 WTR is the absolute best of the best.  This applies to both cumulative numbers as well as individual ships.  Generally, it works when you examine the best players of each ship.  The best players are achieving very high WTR ratings.  However, once you reach tier 10, things change.  

For example, the highest performing Moskva player on the NA server (100 games), who has a 69% win rate and does 110,000+ damage per game is only getting a WTR of 1583.  When you reduce the qualifying battles to 25, you get someone with an 84% win rate doing 128,000 damage and a WTR of 1784.  That just seems silly.  The same thing is taking place amongst all of the tier 10 ships, but is most prominent with regard to cruisers and battleships.  As another example, the best Yamato driver on the NA server has an 85% win rate and does 155,000 damage per game, yet his WTR is only 1882 in the Yamato.

I think the WTR rating also needs to be reevaluated when it comes to aircraft carriers.  With the recent AA buffs and the constant nerfs to CV's, people who are achieving excellent results in the current meta aren't being rewarded in terms of WTR rating due to the massively inflated damaged numbers achieved by CV driver in the early release of the game.

I love the work you have put into the rating, but at tier 10, I think the formula needs to be adjusted.

 

Share this post


Link to post
Share on other sites
6 hours ago, Pope_Shizzle said:

Something about the WTR calculations at tier 10 doesn't feel right to me.  I think that everyone would agree that achieving a 2000 WTR is the absolute best of the best.  This applies to both cumulative numbers as well as individual ships.  Generally, it works when you examine the best players of each ship.  The best players are achieving very high WTR ratings.  However, once you reach tier 10, things change.  

For example, the highest performing Moskva player on the NA server (100 games), who has a 69% win rate and does 110,000+ damage per game is only getting a WTR of 1583.  When you reduce the qualifying battles to 25, you get someone with an 84% win rate doing 128,000 damage and a WTR of 1784.  That just seems silly.  The same thing is taking place amongst all of the tier 10 ships, but is most prominent with regard to cruisers and battleships.  As another example, the best Yamato driver on the NA server has an 85% win rate and does 155,000 damage per game, yet his WTR is only 1882 in the Yamato.

I think the WTR rating also needs to be reevaluated when it comes to aircraft carriers.  With the recent AA buffs and the constant nerfs to CV's, people who are achieving excellent results in the current meta aren't being rewarded in terms of WTR rating due to the massively inflated damaged numbers achieved by CV driver in the early release of the game.

I love the work you have put into the rating, but at tier 10, I think the formula needs to be adjusted.

 

I totally agree. Averaging almost 100K damage and more than 60% solo wins in the Zao, that's barely blue. Green I think. Feels bad man.

Share this post


Link to post
Share on other sites
9 hours ago, Pope_Shizzle said:

I love the work you have put into the rating, but at tier 10, I think the formula needs to be adjusted.

I'm pretty sure this is because of "damage over expected".  It's easy to do 200% or more over expected damage in whatever low-tier ship you want to pick, and pull 2000-2400 WTR all day long.  Grab a buddy and make a seal-clubber division and increase your WTR simply by winning games.  Here's some comparison between all of my tier 10s, along with my top WTR ships with at least 25 games.  I've added my server WTR rank and my expected/done damage ratio.

2ey84xx.jpg

vsm1i8.jpg

I'm currently the top NA player with 25 games in both the Des Moines and the Gnevny, with an 800 point WTR gap between the two.  I think the numbers on the right illustrate why, although I'm sure the other stats are partially to blame as well.

It's simply easier to blow past expected values for lower-tier ships.  The "ceiling" is very high when you're playing against learning players, and the "expected" value is depressed by those players.  For high tier ships, the "expected" value is much closer to the "ceiling", so it's much more difficult to deal 200% or higher, and it appears that WTR is still reflecting that.

Share this post


Link to post
Share on other sites
2 hours ago, Mesrith said:

I'm pretty sure this is because of "damage over expected".  It's easy to do 200% or more over expected damage in whatever low-tier ship you want to pick, and pull 2000-2400 WTR all day long.  Grab a buddy and make a seal-clubber division and increase your WTR simply by winning games.  Here's some comparison between all of my tier 10s, along with my top WTR ships with at least 25 games.  I've added my server WTR rank and my expected/done damage ratio.

2ey84xx.jpg

vsm1i8.jpg

I'm currently the top NA player with 25 games in both the Des Moines and the Gnevny, with an 800 point WTR gap between the two.  I think the numbers on the right illustrate why, although I'm sure the other stats are partially to blame as well.

It's simply easier to blow past expected values for lower-tier ships.  The "ceiling" is very high when you're playing against learning players, and the "expected" value is depressed by those players.  For high tier ships, the "expected" value is much closer to the "ceiling", so it's much more difficult to deal 200% or higher, and it appears that WTR is still reflecting that.

Kind of reminds me of WN7 in WoT. It was so easy to pad with mid tier tanks.

Share this post


Link to post
Share on other sites
5 hours ago, Mesrith said:

It's simply easier to blow past expected values for lower-tier ships.

I wish they wouldn't call them expected values when I spent so many posts explaining that WN8's expected values are not averages. The whole point of expected values is to adjust for population skill  :kjugh:

Warships Today's rating relies on the 5%*(tier-5) adjustment to handle the tier skill escalation, which is roughly half the magnitude it should be. It also won't correct for ships within a tier, such as Kongo vs ARP Kongo or Murmansk vs Omaha.

There are also skill-scaling problems, where it's much harder to double the damage output of a mediocre player in a BB than a DD, but people put up with that in WN8 for years.

Share this post


Link to post
Share on other sites
7 hours ago, RichardNixon said:

I wish they wouldn't call them expected values when I spent so many posts explaining that WN8's expected values are not averages. The whole point of expected values is to adjust for population skill  :kjugh:

Warships Today's rating relies on the 5%*(tier-5) adjustment to handle the tier skill escalation, which is roughly half the magnitude it should be. It also won't correct for ships within a tier, such as Kongo vs ARP Kongo or Murmansk vs Omaha.

There are also skill-scaling problems, where it's much harder to double the damage output of a mediocre player in a BB than a DD, but people put up with that in WN8 for years.

I agree that terminology can be debated. :)

What percentile do the WN8 "expected" values correspond to?

 

 

 

OK, I agree there is a problem, let's do something about this.

I still have a better fix in mind, but for now let's tweak the parameters of the same formula.

How about:

"NeutralLevel = 7.5" (was 5.0)

"PerLevelBonus = 0.10" (was 0.05)

"ShipsFragsImportanceWeight = 10" (was 5) 

"WinsWeight = 0.20" (was 0.25)

"FragsWeight = 0.30" (was 0.20)

Share this post


Link to post
Share on other sites
Just now, WarshipsToday said:

I agree that terminology can be debated. :)

What percentile do the WN8 "expected" values correspond to?

WN8 expected values run pretty close to tier 10 average skill, which is far above 50th centile for most lower tier tanks.

You can't just pick a centile and use that for each tank/ship, although if you aim really high (say top 0.1%) then it's better than using the average. xTE uses that technique. Still sucks for unpopular low-tier tanks unless people switch to padding them.

WN8's a regression method. You compare each player's overall performance to their individual tank performances, and so the final result is population-independent. As a (very) primitive example, imagine calculating ship winrate / overall winrate for each player's ship, and then averaging the results per ship.

Share this post


Link to post
Share on other sites

@WarshipsToday So, I got the "na.warships.today" domain whitelisted for images on the NA forums a few days ago (see here: http://forum.worldofwarships.com/index.php?/topic/80849-any-groups-focused-on-teachingimprovement-still-around/page__pid__1980565#entry1980565 ), but the images do not work in signatures due to an "unable to calculate size" error.  No clue what would cause this, guessing it's a problem with improper metadata in the picture?  If it was the whitelist blocking them, it would give something like "you've linked to a site that the administrator does not allow links to."  I'm guessing uploading to imgur fixes the metadata on the picture, thus they can be posted then.

Share this post


Link to post
Share on other sites
3 hours ago, Tedster59 said:

@WarshipsToday So, I got the "na.warships.today" domain whitelisted for images on the NA forums a few days ago (see here: http://forum.worldofwarships.com/index.php?/topic/80849-any-groups-focused-on-teachingimprovement-still-around/page__pid__1980565#entry1980565 ), but the images do not work in signatures due to an "unable to calculate size" error.  No clue what would cause this, guessing it's a problem with improper metadata in the picture?  If it was the whitelist blocking them, it would give something like "you've linked to a site that the administrator does not allow links to."  I'm guessing uploading to imgur fixes the metadata on the picture, thus they can be posted then.

Thanks for getting it on the white-list.

If you rehost on imgur then it works.
If you then compare the two files (from imgur & from warships.today) they are identical (on binary level).
So I think the files as such are fine.

It could be some HTTP headers; we have experimented with quite a few but nothing helped. 

We're a bit out of ideas at this point, unfortunately :(.

Share this post


Link to post
Share on other sites
On 5/30/2016 at 0:20 AM, WarshipsToday said:

I agree that terminology can be debated. :)

What percentile do the WN8 "expected" values correspond to?

 

 

 

OK, I agree there is a problem, let's do something about this.

I still have a better fix in mind, but for now let's tweak the parameters of the same formula.

How about:

"NeutralLevel = 7.5" (was 5.0)

"PerLevelBonus = 0.10" (was 0.05)

"ShipsFragsImportanceWeight = 10" (was 5) 

"WinsWeight = 0.20" (was 0.25)

"FragsWeight = 0.30" (was 0.20)

I think you're moving in the right direction.  However, I think you value kills too highly.  The old value seemed ok.  The perlevel bonus is a good change.  I think that the Neutral Level is way too high.  

 

I wonder if its possible to (instead of having a neutral level), to have low tier games contribute less to someone's overall WTR, but allow them to keep high ratings on a per ship basis.  If someone is the best Umikaze player in the world, he shoudl be recognized as such.  But the fact that he drives a tier 2 well should have significantly less impact on his overall score than someone who drives tier 6's well

 

Does that make any sense?  I concede that the theorycrafting involved with the WN calcuations are a little beyond me, but the formula you looks to be using seems significantly less complex.

Share this post


Link to post
Share on other sites
2 minutes ago, Pope_Shizzle said:

I think you're moving in the right direction.  However, I think you value kills too highly.  The old value seemed ok.  The perlevel bonus is a good change.  I think that the Neutral Level is way too high. 

In a game with hugely variable ship HP within a tier, frags may well be a stronger correlator than damage. If you weren't worried about padding (shot-holding), the best value would probably be higher.

Neutral level just moves the scale around. Increasing it by one tier does the same thing as reducing everyone's WTR by 5% (or 10% with the new values).

 

10 minutes ago, Pope_Shizzle said:

I wonder if its possible to (instead of having a neutral level), to have low tier games contribute less to someone's overall WTR, but allow them to keep high ratings on a per ship basis.  If someone is the best Umikaze player in the world, he shoudl be recognized as such.  But the fact that he drives a tier 2 well should have significantly less impact on his overall score than someone who drives tier 6's well

You can do that, technically, but there's still a strong sealclubbing incentive. Play only low tiers and you'll get a much better rating than someone who plays mostly high tiers. Tier weighting is more interesting as an addition to tier scaling than as a replacement for it.

 

Share this post


Link to post
Share on other sites

At some point would you mind sharing some basic server numbers that I assume you can access?

How many players on the NA server have:

  • Created an account and played a game?
  • Played 100 games or more?
  • Reached tier 5?
  • Can you see how many of each tier 10 ship exist?  Do 1000 players even have a Yamato, for example?

Our nightly population doesn't really give a clear picture of how many people are actually playing this game.

Share this post


Link to post
Share on other sites
Just now, Tedster59 said:

minor thing, but can you add a "Kill/Death" column to the individual ship stats?

Which page/table do you mean this in?

Technically, I can add them no problem. But some tables become very large that way and break layout / mobile friendliness. I haven't thought of a good solution.

Share this post


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...