Jump to content


WoT-sup - An ELO-like rating system for player strength

stats ELO rating skill

  • This topic is locked This topic is locked
8 replies to this topic

Erapin #1 Posted 20 May 2013 - 08:16 AM

    Sergeant

  • Player
  • 20639 battles
  • 254
  • [HO_PO] HO_PO
  • Member since:
    05-11-2012
It has come to my knowledge that my old thread has been archived, hence the new posting. Here goes!

WoT-sup - Statistical User Proficiency

What new?
For the resident conspiracy theorist, I added stats about MM tier assignment. It's a report that'll tell you how often a player is assigned to top, mid and bottom tiers. Are you a VICTIM? Find out! ;)

What is it?
WoT-sup is a website that attempts to rate players proficiency (or play strength) in a statistically sound way. Users submit data in the form of battle result files to the site and receive a rating - As a side-effect, all their allies and opponents are also given a rating.

How does the rating work?
The rating system is a modification of the Glicko rating system (that works a bit like ELO), with allowances made for multiplayer teams and different tank abilities. In short, your rating is calculated as a function of your performance over time, with account taken to the proficiency of your own teams and the opponent teams and more importance placed on recent matches.
A win against a stronger team gives a bigger gain, a loss against a weaker team gives a bigger drop. Conversely, a win against a weaker team gives a marginal gain and a loss against a stronger team gives a marginal drop.

What makes that so different from win/loss, efficiency ect. etc.
There are several differences, but the most fundamental one is that the WoT-sup rating is based only on wins.
The calculation method is complicated, true. The big upside is that this is a rating system that can be used in a proficiency based match-making system (skill tiered MM, league play etc.). In a stat-based system, proficiency based matching will make your W/L tend towards 49%, regardless of your skill. In league play, efficiency will tend to be depressed to an average value if you are average in your own league. Also - There is no real way to compare the strength of players across leagues.

How do I try it out?
Simple. Go to http://www.atonet.se...ic/download.php and download the client (data committer) and run it as you play. It will send result files to the WoT-sup server for rating. Ratings are calculated every night.
The client can also work a bit like the XVM plugin, giving you a report on the rating of the players participating in your current battle. Usually, the higher tier you play, the more precise the ratings will be.

More detailed info?
Certainly! This wall of text is long enough as it is. More can be had here.
Feedback happily accepted!

Changelog
2013-02-19: New feature on the server: Report on recent treatment by the MM, tier assignment.
2013-02-18: OpenID login with Wargaming credentials.
2013-02-13: New client version with timed updated on ratings report.
2013-01-02: New win rate stats on the player page.
2012-12-21: New feature in the client: Battle briefing, XVM style.
2012-11-23: Implemented "last met player" on player page - Will show team, time, map and vehicles when the players being compared last met.
2012-11-22: Implemented "recent ratings" on player page - Will show snapshot ratings for the last week.
2012-11-17: Fixed import and rating history graph.
2012-11-15: New import version - New client, eradicated submitter bias and more.

Edited by Erapin, 20 May 2013 - 08:16 AM.


UM8D #2 Posted 21 May 2013 - 06:47 AM

    Warrant Officer

  • Player
  • 29423 battles
  • 595
  • Member since:
    09-23-2011

View PostErapin, on 20 May 2013 - 08:16 AM, said:




For the resident conspiracy theorist, I added stats about MM tier assignment. It's a report that'll tell you how often a player is assigned to top, mid and bottom tiers. Are you a VICTIM? Find out! ;)


How can you tell that?



General:
How is it decided which team is 1 and 2?

regards

Erapin #3 Posted 21 May 2013 - 03:35 PM

    Sergeant

  • Player
  • 20639 battles
  • 254
  • [HO_PO] HO_PO
  • Member since:
    05-11-2012

View PostDiola, on 21 May 2013 - 06:47 AM, said:

How can you tell that?

The result files contain information on who drives what vehicle. Calculating top/bottom tiers from that is trivial.

View PostDiola, on 21 May 2013 - 06:47 AM, said:

General:How is it decided which team is 1 and 2?regards

Map dependent. Before, the maps used to be marked as "I" or "II", but that's gone now, meaning it's a bit harder to know...

UM8D #4 Posted 21 May 2013 - 10:53 PM

    Warrant Officer

  • Player
  • 29423 battles
  • 595
  • Member since:
    09-23-2011

View PostErapin, on 21 May 2013 - 03:35 PM, said:

The result files contain information on who drives what vehicle. Calculating top/bottom tiers from that is trivial.

I do not understand. But whatever. It is probably because of the language and the Google translator. (I'm German)



View PostErapin, on 21 May 2013 - 03:35 PM, said:

Map dependent. Before, the maps used to be marked as "I" or "II", but that's gone now, meaning it's a bit harder to know...

Can we say that a particular side of a map is always the same team (1 or 2)?



I hope google does something intelligible out of it here. :)

tango_delta #5 Posted 22 May 2013 - 04:54 AM

    Captain

  • Player
  • 11465 battles
  • 2,026
  • Member since:
    03-17-2011

View PostErapin, on 20 May 2013 - 08:16 AM, said:

What new?
For the resident conspiracy theorist, I added stats about MM tier assignment. It's a report that'll tell you how often a player is assigned to top, mid and bottom tiers. Are you a VICTIM? Find out! ;)
Sounds really nice must test.

Have you ever thought about doing a luck rating? Something like I've outlined here:
http://forum.worldof...99#entry4825599
This could be done from replays.

Erapin #6 Posted 23 May 2013 - 09:47 PM

    Sergeant

  • Player
  • 20639 battles
  • 254
  • [HO_PO] HO_PO
  • Member since:
    05-11-2012

View PostDiola, on 21 May 2013 - 10:53 PM, said:

Can we say that a particular side of a map is always the same team (1 or 2)?I hope google does something intelligible out of it here. :)

Intelligible or not, I'll try to answer! :)

We know that the same starting positions receive the same number in each match. A clear indication that this is in fact true, is that the "Map statistics" page show bias for one starting position over another across thousands of games. If the numbers for the starting positions had been randomly allocated, every map would have a 50/50 win ratio.

UM8D #7 Posted 24 May 2013 - 02:18 AM

    Warrant Officer

  • Player
  • 29423 battles
  • 595
  • Member since:
    09-23-2011

View PostErapin, on 23 May 2013 - 09:47 PM, said:

Intelligible or not, I'll try to answer! :)

We know that the same starting positions receive the same number in each match. A clear indication that this is in fact true, is that the "Map statistics" page show bias for one starting position over another across thousands of games. If the numbers for the starting positions had been randomly allocated, every map would have a 50/50 win ratio.

hihi, thx :)

UM8D #8 Posted 26 May 2013 - 06:17 AM

    Warrant Officer

  • Player
  • 29423 battles
  • 595
  • Member since:
    09-23-2011

View PostErapin, on 20 May 2013 - 08:16 AM, said:

For the resident conspiracy theorist, I added stats about MM tier assignment. It's a report that'll tell you how often a player is assigned to top, mid and bottom tiers. Are you a VICTIM?

I think absolut:(Is it this?)
Posted Image
Edit: Today:
Posted Image

http://www.atonet.se...0b-ae5572c059dc

Edited by Diola, 05 June 2013 - 04:53 PM.


VeryRisky #9 Posted 26 May 2013 - 09:02 AM

    General

  • Player
  • 17949 battles
  • 8,694
  • [WJDE] WJDE
  • Member since:
    10-11-2012
I'll give it a go for a bit and see what I find.  Now that's three monitor programs I have running while I play!





Also tagged with stats, ELO, rating, skill

1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users