Sign in to follow this  
zerbah

[eSport] Tournament rating and Swiss system FAQ

 

Frequently asked questions about tournament rating

 

1. How can I check my tournament rating on the https://tss.warthunder.com?

 

To check your tournament rating, you need to log in on the https://tss.warthunder.com website:

login_en.jpg

login_2_en.jpg

After you are logged in, click on the “RATING” button:

rating_button_en.jpg

The tournament rating page will open. The page contains rating leaderboards in different types of tournaments (aviation, ground vehicles, fleet and mixed/combined). These are divided into modes - Arcade Battles (AB), Realistic Battles (RB) and Simulator Battles (SB):

rating_page_en.jpg

Your place in the rating leaderboards will be highlighted if you have already participated in a certain tournament type and mode:If you haven’t participated in a tournament type or mode before, your place in the rating leaderboard for these tournaments will not be shown. For example - if you only played in AB and RB tank tournaments, your place in rating leaderboards will be shown in “GROUND VEHICLES” section of tournament types and in “ARMY AB” and “ARMY RB” mode leaderboards.

my_raitng_en.jpg

 

2. How does the tournament rating work (how is rating calculated)?

 

Tournament rating on https://tss.warthunder.com is based on the Elo rating system.

 

Each player, who participates for the first time in a certain tournament type and mode, receives the starting rating of 1000 points. This rating will not show in the leaderboard until player plays at least one match in this tournament type and mode.


Tournament rating is calculated by results of each tournament match in the following way:

 

We have two players in a match, player A and player B.

 

In order to calculate rating after the match for player A, first we have to calculate the expected score for player A in a match with player B is:

 

Expected_score_main.gif

 

Where:

ЕА - mathematically calculated expected score of player A in a match against player B (has value from 0 to 1)

RA - current rating of player А

RB - current rating of player B

 

Next, we calculated the number of points that player A will have after the match:

 

Rating_calc.gif

 

Where:

R’A - rating of player A after the match with player B

RA - current rating of player A

SA - number of points received based on match results: win - 1, draw - 0.5, loss - 0

K - coefficient, value of which is fixed and equals to 10.

 

Example of rating calculation:

 

Before the match, the player A has the current rating of 1225 points and player B has the current rating of 925 points.

 

The expected score of player A will be:

 

Expected_score_A.gif

 

The expected score of player B will be:

 

Expected_score_B.gif

 

If the player A wins the match, the resulting ratings will be:

 

5.png

 

6.png

 

In case that the player B wins the match, the resulting ratings will be:

 

7.png

 

8.png

 

Based on these calculations, it is clear that player with higher rating receives less points for winning and loses more points for losing a match against the player with low rating.

 

3. How is rating in team tournaments calculated?

 

Rating for players in team tournaments is calculated individually for each player in the team by match results. The current rating for each player before the match is based on the average rating of all players in the team.

 

4. How can I participate in rating tournaments?

 

In order to participate in rating tournament, you need to have the rating that is in the range of the allowed rating of that tournament.

 

To check if you can participate in a rating tournament, move your mouse to the rating icon of the tournament in the list. If the rating allows you to participate, you will be able to press the “JOIN” button of this tournament and register your team.

 

In case that your rating is not in this tournament rating range, you will see the following message:

join_rating_tournament_en.jpg

 

5. How often is the rating updated on the website?

 

Rating is updated daily on the website.


6. How can I check my opponent’s rating for the next match?

 

You can check your opponent’s rating by clicking on his team in the tournament bracket or in the teams list:

opponnent_rating_en.jpg

 

Frequently asked questions about the Swiss system tournament bracket

 

What is Swiss system tournament bracket?

 

Swiss system is one way of organizing sport events. The main advantage of this system is that tournaments are played without teams being knocked out from the competition.

 

In the first round, the matching pairs are created randomly: teams are separated in two groups after which the first team from the first group is matched with the first team from the second group, the second team from the first group is matched with the second team from the second group and so on. Example: if there are 30 teams in the tournament, the 1st team will play against the 16th team, the 2nd team will play against the 17th team, etc.

 

If the number of teams is not equal, the team that has the last number will receive a point (win) in the first round without playing a match.

 

Starting from the 2nd round, the teams are separated into groups with equal numbers of received points. Based on the first round, there will be 3 groups - teams who won, those who played a draw and those who lost. If there is an non equal number of teams in one of the groups, one of the teams is moved to the closest group based on points.

 

For the next round the matching pairs are formed in the same way, but this time from the teams of the same group -  the strongest team from the top of the group is playing (if possible) against the strongest team of the second half of the same group. Finally, there is a rule that any two teams cannot play more than one match against each other during the whole tournament.

 

Number of rounds is calculated by the following formula:

 

[log2 N] + [log2(k–1)]

 

Where:

N — number of teams

k —  number of prize places (where k=1, [log2(k–1)]=0).

As an example, in order to find a winner in a tournament for 8 teams, 3 rounds will be enough, if there are 16 teams - 4 rounds. To decide the 4 best teams, a tournament for 8 would need 4 rounds, while a tournament for 16 teams would need 5 rounds. Places in the tournament are distributed based on the number of points received in matches (win - 1 point, draw - 0.5 points, loss - 0 points).

 

If some teams have the same number of points at the end of tournament, their places are distributed based on Buchholz score. When used as an alternate scoring system, each team's Buchholz score is calculated by adding the raw scores of each of the opponents they played and multiplying this total by the team's raw score. The idea is that the team who played against more difficult opponents receives higher place in the tournament.

 

An example of the Swiss system tournament bracket:

swiss_bracket_en.jpg

  • Upvote 2

Share this post


Link to post
Share on other sites
33 minutes ago, Jaxacus said:

Question. Do the tournament ratings get reset after each season?

For now no.

medal medal medal medal medal medal

Share this post


Link to post
Share on other sites
On 14/03/2019 at 13:16, zerbah said:

 

...

 

If some teams have the same number of points at the end of tournament, their places are distributed based on Buchholz score. When used as an alternate scoring system, each team's Buchholz score is calculated by adding the raw scores of each of the opponents they played and multiplying this total by the team's raw score. The idea is that the team who played against more difficult opponents receives higher place in the tournament. 

This part is actually wrong for higher-placed teams, which is perfectly illustrated by just ended RB air race tournament. To be more precise - "the idea" does not work for teams which have, at the end of the tournament, 1 loss each (and zero draws, so same amount of points). Which often are most or all of top-3 teams in a tournament, and sometimes as many as top-5 even. As i will explain below, for all such teams, Buchholz score is just _random_. I.e., as implemented, Buchholz score often does not anyhow depend on strength of teams fought by any such "just 1 loss in a tournament" team.

 

Surprised? Read on!

 

Consider the following facts which happened in the above mentioned RB race tournament today:

 

- Jaxacus (team "Bumblebee") had 1 loss and took 1st place, with Buchholz score = 0.55; i also had 1 loss and took 2nd place with Bychholz score = 0.52. But:

- cumulative (summed up) rating of pilots Jaxacus defeated = 4190sh, but cumulative (summed up) rating of pilots i defeated = 6000+;

- in Jaxacus vs me race ("direct face-off" so to say) - he lost, which directly indicates i was stronger pilot than him, during this tournament;

- he lost (crashed) 2 planes during the tournament, while i lost 0.

 

So by all sensible indications, i was stronger than Jaxacus in this particular tournament, but Buchholz coefficient says otherwise. Why?

 

It's because "whomever lost earlier = will have lower Buchholz score" among all teams which end up having just 1 loss during a tournament, and yet it is quite random which of those teams will encounter actually one of the strongest teams of the tournament (and lose to that one) earlier than others. Which is how Buchholz ends up being actually _random_ for such teams (again, 1 loss per tournament ones).

 

This is perfectly indicated by above mentioned RB Air Race tournament which just ended.

 

In round 2 of the tournament, i did my loss vs Rider2 (team "Potato"), who's one of strongest air racers in WT for years. While Jaxacus in that same round 2 - won vs Oreska, who - no offense - is very weak racing pilot, as evident from very results of their race: Jaxacus lap time = 4:49, Oreska's lap time = 8:03. When round 2 races were set up, all 4 those teams had 2 points (for winning in round 1, as evident from tournament bracket of this tournament, available at http://tss.warthunder.ru/index.php?action=tournament&id=8711# ). Therefore, the system considered all those 4 racers "same strength" for round 2 - since all 4 had same number of points. But, as just explained, actual strength of those 4 pilots was not the same: 3 of them were strong racers (me, Jaxacus, Rider2 - as evident from the tournament's results), while Oreska was weak in both actual playing strength (see lap times just above), as well as in terms of personal rating (0 at the time), and not surprisingly placed much lower (7th) in final results, as well.

 

And that's where this randomness comes from: in early rounds, the system will inevitably assign "actually strong teams" to some matches, "strong and weak" for other matches, "weak and weak" for yet others. That's how some weak teams go as "winners" to round 2 and even round 3 - from the latter kind of round 1 pairs (weak vs weak). And then "some unlucky team" will have an "early loss", akin to the loss in round 2 of this RB Air Race tournament i just had. Then, by rules of the Swiss system, that unlucky team will be "moved down" to play its following rounds vs other teams who _also_ "have 1 loss already". Naturally, those teams, as opponents, have _lower_ score than "no losses yet" teams which continue to duke it out in "winners branch" in mid-rounds of the tournament - just like Jaxacus did in this RB air race till round 6, which is when we met and i defeated him. And there we go: this "unlucky team" (just like i was in this RB tournament) - could actually be better than higher-Buchholz teams in practice, but nonetheless it gets lower Buchholz, only for having that 1 loss earlier, during the tournament, than other teams who had the same number of points at the end of the tournament.

 

The rule which makes it happen that way, i.e. the rule which _guarantees_ that "earlier loss = lower Buchholz score" - is this one, quote (from the tournament's rules): "In the following rounds, all the players in the bracket are divided in the groups of the same amount of points that they have received in their matches".

 

How is it fair that "earlier loss" means lower Buchholz score no matter what, for all teams which happened to have just 1 loss in a whole tournament? It is not. And so, it needs to be fixed. Otherwise we'll continue to have pretty many tournaments where "winner" is actually decided not by strength of leading teams, - but merely by random "dices" rolled when pairing up teams in early rounds of a tournament.

 

How can this be fixed? Well, here's what i propose. Per all the above, to replace the part quoted in this post - by the following rule:

 

"If some teams have the same number of points at the end of tournament, their places are distributed based on which team have lost less vehicles. If both number of points and number of lost vehicles is the same, then all such teams are placed on the average place of all such teams, rounded down. Example: if 3 teams have same number of points and same number of lost vehicles, and have enough points to occupy places #7, #8 and #9 - then all three will be placed 8th".

 

 

P.S. All that said, i do not request to anyhow change the results of this RB Air Race tournament used as an example for this post; i wrote all this because the system needs to be fixed, no other reason. Jaxacus very well earned his win as far as i am concerned, doing best he can. Well done, Jaxacus! :) o7

Edited by Fins_FinsT
medal medal medal medal medal medal medal medal

Share this post


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

...

will respond when i get sober

 

sorry if weird post, felt cute, might delete when sober :attack: xD

1. I am racing in WT since 2014, top level. More than a half racing tournaments and events through the years - were won by me. What you said is thus just... funny, to me. ;)

2. I understand how it works. Proof: (1+5+1+5+3+6+5) * 6 / 300 = 0.52 (that's raw scores of all teams i faced in said tournament, multiplied by my raw score, divided by (probably just static coef) 300 - which gives exactly 0.52 as we see in the table. Can't be a coincedence, given that's how Buchholz works in general.

3. I am 100% sure, because you know - math (see p.2 just above).

 

And you probably have no idea that the problem i described in my above post - is actually well known problem of Buchholz tie-breaker systems. Quote: "The major criticism of this system is that tie-break scores can be distorted by the set of opponents that each player plays (especially in early rounds)". Source.

 

What i did above - is merely demonstrating in detail, with a specific WT tournament example, how this problem manifests itself in WT tournament system.

 

I'd love to discuss this further, sir, but perhaps better when you do get sober? Respectfully.

 

 

 

Edit: further data, for sober times. Here's more examples of recent WT tournaments with swiss system where "early loss" or two - would result in Buchholz score way lower than generally adequate if comparing other parameters (kills, vehicles lost), because teams were dropping into "losers branches" early on:

 

http://tss.warthunder.ru/index.php?action=tournament&id=8548#info-info_bracket : tied score for places 2 to 5, 2 losses each of those 4 teams; but, team EMP has dramatically lower Buchholz score than other three teams, for team EMP had two losses rather early - in rounds #2 and #4. Both times fighting vs one of top-5 teams of the tournament ("strong-strong" matches early in a tournament produce a loser and are not avoidable by teams, as they do not choose whom to fight). Notice how team EMP has markedly less lost vehicles than 3 other teams. It'd be 2nd if using my proposition, and well deserved 2nd - but Buchholz score puts it to 5th;

 

http://tss.warthunder.ru/index.php?action=tournament&id=8610#info-info_bracket : team 16V_747 placed 6th, tied score with 4 other teams, all 5 those teams have 2 losses. Note how this team loses in 1st round and has dramatically lower Buchholz score than the rest (2nd...5th placed teams). By my proposition, it'd take 2nd place in the tournament instead, because of less lost vehicles (than other 4 teams). Also, note how team SoSoLucku in this same tournament, placed 2nd, - has highest Buchholz score of those 5 teams, and how it lost very late (rounds 4 and 6, out of 6);

 

http://tss.warthunder.ru/index.php?action=tournament&id=8621#info-info_bracket : team ZEBRALLIOT loses in 1st round (vs other strong team) and ends up with much lower Buchholz score (than other two teams with 2 losses) at the end, thus taking 4th place. This team would be 3rd instead, if tie-breaking by vehicles lost;

 

http://tss.warthunder.ru/index.php?action=tournament&id=8686#info-info_bracket : team bb lost in 1st and 4th rounds, while team Queen lost in 2nd and 5th (so at later points); the former has markely lower Buchholz score and is placed 4th, the latter placed 3rd - while by vehicles lost it'd be the other way around;

 

http://tss.warthunder.ru/index.php?action=tournament&id=8609#info-info_bracket : team _123 (4th place) has much lower Buchholz than team TANK (3rd place), for losing earlier rounds it did. While by vehicles lost, team _123 is clearly 3rd place.

 

Those are all tournaments during last week, only. And i did not check 'em all (far from). Thoughts?

Edited by Fins_FinsT
medal medal medal medal medal medal medal medal

Share this post


Link to post
Share on other sites
On 03/06/2019 at 14:51, Ukiii said:

http://tss.warthunder.ru/index.php?action=tournament&id=8711#info-info_bracket

u played vs

TH 2pts

Potato 10pts

equisde 2pts

123 10pts

wt 6pts

Bumblebee 12pts

PoweRBull_ 10pts

what gives 52 - (0.52) next to YOUR points, idk why its shown like that but its ok

 

Bumblebee played vs

asd 0pts

oreska 8pts

pipi 10pts

Fizzi 5pts

PoweRBull_ 10pts

Fins 12pts

Potato 10pts

=55 (0.55)

thank me later for destroying your childhood :fighter:

it aint perfect but prob little bit better than k/d since in some games enemies might not spawn all vehicles or just kill themselves which would give u less k/d, now at least it looks at who played vs stronger teams in that tourney

 

maybe if u lose early u get weaker enemies later which also gives u less (0.00) score but at the end of the day u have no control over whom u gonna face so just play, win, and if u lose cuz of that score - deal with it, other team faced higher placed opponents in that tourney

 

its just a small weekend tourney which will not be played endlessly just to see who is really better, for the big tourneys there is a different system

You destroyed nothing, as math i shown in my previous post works just as well as yours, - for my, Bumblebee's or any other pilot's Buchholz score. Like i said, i knew how it is calculated in principle.

 

Now would you please make an effort for me? I ask you to think hard about the following, and tell me if you understand my point. Pretty please!..

 

The problem is not that there is a mistake in how Buchholz score was / is calculated, sir. The problem is that in described by me conditions, the _proper_ calculation - is unfair.

 

Let's try to get to the point using your way of describing things. Do you see how i got only "2 pts" in 3rd round because i played vs a weak opponent? He was weak, de-facto, in this particular tournament, because he only made 2 pts whole tournament. I hope you see that. 3rd round "gave" team Bumblebee whole "+8 pts" to his Buchholz score in compare to me, so this is "critical moment" in this particular example.

 

Now please tell me: why, exactly, did i play against a weak opponent in 3rd round - while Bumblebee played vs a strong one (10 pts) in that same 3rd round? Could it be much caused by the fact that i had 1 loss "already" when the system was making pairs for round 3, while Bumblebee had 0 losses at the time?

 

I'd be very surprised if you'd answer "no" to the latter question, given the particular rule i quoted in my very 1st post in this topic.

 

So then, if you'd look to round 2 where i got a loss which trigged whole thing, - you will see that i had much higher chances to lose in round 2, because i was paired against one of strongest teams (Potato, 10 pts), while team Bumblebee was paired against Oreska (who's much weaker than his 8 pts score suggests, as explained by lap times in one of my previous posts). Now, do you understand that those specific pairs in round 2 - were made completely random? And if you do - do you understand _why_ it was completely random pairings in that round 2?

 

The whole deal is obviously a subject to fluctuations here and there, of course; like only 5 pts for team Bumblebee in round 4. But on _average_, it is absolutely mathematically obvious that "much earlier losing" for a team which ends with just 1 loss whole tournament - will mean "much lower Buchholz score". Period. Do you now understand why this is so? I hope you do.

 

If you still do not, i can describe what's going on "under the hood" of Swiss tournament system in yet greater detail, if you'd be interested.

 

Because once you realize my point, i think you'd wholeheartedly agree things often can't be fair about this Buchholz tie-breaker approach - and thus better solution is desirable.

 

P.S. Oh almost forgot. You seem to still think i can't "deal with it" about me ending 2nd in that particular tournament. I already told you why i couldn't care less - i am winning lots of things (and still losing some) for _years_. Any "ego" i could possibly have about those competitions - is long time ago fulfilled. Completely fulfilled. I made that P.S. in my 1st post to clear that up, too. How else can i confirm that i write here for the good of _all_, trying to improve the scoring system, trying to remove built-in circumstantial injustice of Buchholz tie-breaker (which other examples i gave in my previous post so clearly demonstrate)? It kinda insults me, even, that you suspect i made this whole exchange only because i "lost" as in not taking 1st place in that RB race tourney. It kinda means you suspect me a liar, given P.S. of my 1st post in this topic. Not nice, you know... So, please tell me what you want me do for you to believe i ain't doing this for my own "feelings" and/or "interests", - and if possible, i'll just do it for you.

Edited by Fins_FinsT
medal medal medal medal medal medal medal medal

Share this post


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

http://tss.warthunder.com/index.php?action=tournament&id=8752#info-info_bracket

BbIDLO and Leet

it aint the best example cuz they both have 1 loss in matches against each other but BbIDLO lost in 2nd round and still got 1st over Leet that lost in 6th

 

the point is tho - SW will always be random but old system was random too and since SW is new and wont be changed anytime soon, this is what u have to deal with 

this (0.00) score is prob best atm cuz it looks at who faced harder opponents as oppose to k/d that can be manipulated by teams not showing up, killing themselves or not using all spawns available

 

here u have a chance to fight back, the old SE used for small tourneys didnt allow that

 

Which part of " on _average_ " term in my previous post eludes you? The example you gave - is the exception to the trend, and it is clearly visible why: 1st round. In the 1st round, BblDLO earned 4 pts to his Buchholz score (in your terms), while Leet earned 0 pts. The difference is the result of completely random pairings in 1st round. If those two teams would earn same amount of pts in very 1st round - then they would switch: BblDLO would get 2nd place in the tournament, Leet would become 1st.

 

And, this "1st round Buchholz score of everyone will always be totally random" - is one more bad thing about Buchholz tie-breaking; quite often this "1st round total random difference" for any given pair of players - is even much higher.

 

You ain't even paying attention to what i proposed, man. I never proposed k/d instead of Buchholz score. Will you bother to scroll up to my 1st post and actually read what was proposed?

 

You also did not answer any single question from my previous post.

 

You also still do not understand that me, personally, - i am totally "dealing with" with things as they are. Take a look at the ratings page - i am in top-10 by rating in all 3 difficulty modes. Are you simply UNABLE to comprehend that sometimes people do things not for their personal gain / benefit, but in the attempt to make life of OTHERS better than it is right now?

 

Though feel free to ignore those questions like you ignored earlier ones. It, in itself, tells volumes about your... Attitude. Frankly, i see very little point, if any, to continue this conversation unless you significantly improve said attitude.

 

Adios.

 

P.S. And to whomever would be considering my proposition (in my 1st post in this topic, bold text) properly: the benefit of using lost vehicles (lesser is better) as a tie-breaking for tournaments - is simply that in literally every tournament, vehicles are used and it is possible to lose them (be killed, crash a plane, roll bottom up a tank, etc) - and it is the player(s)' skill only which defines how many were lost. Better teams will always lose less vehicles, on average.

 

Edited by Fins_FinsT
medal medal medal medal medal medal medal medal

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.