Jump to content
WarshipsToday

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

Recommended Posts

Do you see all API at once or can you separate individuals?  I know I had at least 2 caps Yesterday in my Yamato, and I'm showing 0's.

Link to post
Share on other sites
19 minutes ago, map381 said:

Do you see all API at once or can you separate individuals?  I know I had at least 2 caps Yesterday in my Yamato, and I'm showing 0's.

I can split by week, but not by day unless I collect a lot of extra data. I can see that your Yamato has zeroes anyway.

Things I'd need for a bug report:

  • Whether it's some caps & defence or all.
  • If only some, whether this is related to cap type.
  • Whether they show up in the service record.

If there's nothing in the service record then the API is unlikely to be any better.

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

Any progress or expected updates on this?  Even ignoring high-tier ships, there are still some odd comparisons to be found.

 

  • My Furutaka has a WTR of 2198, presumably because of it's 80% win rate in the 10 games I played in it, despite the fact that my 35k damage wouldn't even be in the top 1000 players.  My 80% win rate is 34% higher than expected, but damage is only 14k higher than expected.
  • My Arkansas Beta has a lower WTR of 2145.  My 72% win rate is 22% higher than expected, but my 74k damage is over double the expected value of 30k, and is #2 on the global warshipstats.com leaderboard.

I still don't understand the point of including win rate in a metric that should be grading the other stats in relation to how they affect win rate.  We don't give athletes stat bonuses for winning the game, because that wouldn't make sense.  If we want to see how often someone wins in a ship, we can simply look at pure win rate.  If we want to see how they perform individually in that ship regardless of the battle's outcome, that's what a stat rating should do.

We are making (difficult but nevertheless) progress on some features which we hope will help produce a more accurate (for lack of better term) rating. They should be done in next release and then we will have statistics to think of how to adjust the rating.

It's good that you are reporting the anomalies here as I will visit them and use as test cases later on.

Whether to include win rate is a good question, though somewhat philosophical. I kind of hate win rate, but if we throw it out and also throw out def and cap as they are wrong then we are left with mostly damage and then we may as well just look at that. I think if the goal of the rating is to quickly assess how good a player someone is then we keeping win rate makes sense.

Link to post
Share on other sites

You're just always going to have weird results if you continue to include individual ship win rate.  A case can be made that damage is much more accurate with small sample sizes.  If someone plays 20 games in a ship, they're likely to have a reasonable mix of games (some low-damage games from bad positioning or an early death, some high damage games where all goes right) with an average that makes sense based on their skills.  But within those 20 games, their team could have carried them to a win even if they died in the first 4 minutes, while their team could have lost games when they pulled out a top 10% individual performance.  I believe that damage patterns resolve themselves relatively quickly because it's an individual statistic, while win rate has an element of randomness due to the team nature of the game.  I absolutely believe that win rate is indicative of player skill, but it takes many more games to smooth out the random element.  It can be valuable to look at someone's overall win rate, but I don't think most people will play enough games in each ship for that ship's win rate to be meaningful.  I'm sure we can all look at our profiles and recognize the ships that we did really well in but couldn't seem to win as often as we should, or the ships that we did poorly in but got carried more often than not.

 

If WTR was simply an overall rating then including win rate would make a bit more sense due to the much larger sample size.  But when you assign a WTR to each indidivual ship, and some of those will have fewer than 50 games played, the numbers feel pretty arbitrary.  I think cap and defense should obviously count if they're being properly counted on WG's end.  I think damage-above-expected is the main indicator of player skill.  Even if we agree that damage isn't the primary game-winning stat, I think we could also agree that the players that put themselves into positions to do more damage are also likely to do the hidden things that win games.  While someone could potentially pad a rating by just spamming damage into the wrong target, this should be reflected in a lower win rate due to choosing meaningless damage over shooting what's needed in order to win, like destroyers.

 

TLDR:  If I'm right, you should be able to craft a WTR that uses expected-damage values for individual ships (and eventually cap/defense) but ignores win rate.  It would average out to an overall number for a player that should closely reflect what their overall win rate is.

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

I can split by week, but not by day unless I collect a lot of extra data. I can see that your Yamato has zeroes anyway.

Things I'd need for a bug report:

  • Whether it's some caps & defence or all.
  • If only some, whether this is related to cap type.
  • Whether they show up in the service record.

If there's nothing in the service record then the API is unlikely to be any better.

I'll try and remember to grab screenies for ya

Link to post
Share on other sites

IIRC, this has been related to the domination caps not crediting capture points, only standard capture points (which have all but vanished in WG's attempts to stop draws)

Link to post
Share on other sites
On 11/30/2015, 9:46:21, Mesrith said:

TLDR:  If I'm right, you should be able to craft a WTR that uses expected-damage values for individual ships (and eventually cap/defense) but ignores win rate.  It would average out to an overall number for a player that should closely reflect what their overall win rate is.

I like what you write, and I hear you. We will keep this in mind for the next iteration of WTR.

Link to post
Share on other sites

Another possible improvement I'd like to suggest would be a WTR rating by ship class for each player; at least for your overall numbers if doing so for the dailies doesn't work.

Link to post
Share on other sites

New version deployed.

Changes:

  • High Level Summary added - a brief description of the player on top of the player page
  • Charts per Tier and Type can now be coloured by various stats, such as damage dealt (instead of just win rate as before)
  • Data from WG on capture & defense is unreliable and won't be used for WTR calculation
  • Fixes to ship leaderboards - situations where a player deserves to be on a ship leaderboard but isn't should be much more rare
  • Various minor fixes and stability improvement

Your feedback is much appreciated. If you like the site - please share it with friends!
 

 

 

On 12/6/2015, 3:13:13, CaptBellamy said:

Another possible improvement I'd like to suggest would be a WTR rating by ship class for each player; at least for your overall numbers if doing so for the dailies doesn't work.


Great idea, we will implement this.

Link to post
Share on other sites

Just had a look at it: really nice, I like this feature.

High level overview
  • Mostly plays battleships, especially medium-tier and is excellent in them
  • Rarely finishes damaged enemies
  • Extremely often uses torpedoes
  • Key vehicle - Warspite

Link to post
Share on other sites

Currently there is a page for PVE stats and one for random battle stats.

I would like to have a page for ranked battles, that would be more useful than the PVE page I think.

Just a suggestion, if you are looking for ideas :)

Link to post
Share on other sites
 

Currently there is a page for PVE stats and one for random battle stats.

I would like to have a page for ranked battles, that would be more useful than the PVE page I think.

Just a suggestion, if you are looking for ideas :)

This is next on our TODO list.

Link to post
Share on other sites

meanwhile, how does NA's ability to hide your profile affect the API or however warships.today pulls stats? (I currently have my profile set to "link only")

Link to post
Share on other sites
 

meanwhile, how does NA's ability to hide your profile affect the API or however warships.today pulls stats? (I currently have my profile set to "link only")

Given the rumors of a Warships XVM (i.e literally the worst idea in Warships history) I'm interested in this as well.

Link to post
Share on other sites
 

meanwhile, how does NA's ability to hide your profile affect the API or however warships.today pulls stats? (I currently have my profile set to "link only")

Apparently it doesn't do anything at the moment. I can read your stats just fine on the API.

Link to post
Share on other sites
 

Apparently it doesn't do anything at the moment. I can read your stats just fine on the API.

riiip.  just set it to full private, does that change anything?

Link to post
Share on other sites
 

Given the rumors of a Warships XVM (i.e literally the worst idea in Warships history) I'm interested in this as well.

Having played WoT with XVM for the last three years. And then for last year playing WoWs and more recently AW without it I can honestly say that XVM is one of the the most cancerous things the fans a game have introduced to that games community.

Link to post
Share on other sites
 

Well, all I can say is that there is no way to hide data from a would-be XVM while keeping it available for stats sites. Killing XVMs means killing the stats sites. gg.

Fog of war or even partially hiding players names in battle would disable XVM.

Link to post
Share on other sites
 

riiip.  just set it to full private, does that change anything?

Yes. All I can see now is your account creation, last battle and log-out times. No stats.

This is gonna be a pain for interval data collection.

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...