Posts from On-line rating (discussion) Auto-translated

Let's say, I, Vasya Pupkin, and Pupka Vaskin get together and hold a tournament (or maybe we don't actually hold it, but just write that we played), and we get 500, 400, and 300 points, and we enjoy it. A week later, we hold another tournament.
As soon as loopholes appear in rating manipulation, they will be used.

We all know each other quite well, and I think it will be impossible to organize such a fake tournament. It should be obvious!

Therefore, a large tournament should award many rating points, and a small tournament should award fewer.

Playing in a tournament with 16 or 32 players will result in the same total number of games – there will just be more subgroups. I won't argue with that; I agree with the majority opinion. A big request: more specifics. That is, with numbers, etc. The next season is coming up soon, and we need to solve and think about all this in 2 weeks: how many points 1st, 2nd, and 3rd place receive in a tournament of 8 players, and also the number of points awarded to winners of tournaments with 16, 32, and 64 participants.

32 players - 1000
16 players - 500
8 players - 300

In my opinion, this is too much! Winning a tournament with 16 participants is almost as difficult as winning a tournament with 32. As I wrote above, the number of games is the same, and there are just more subgroups!

I suggest introducing a rating difference limit! That is, if the rating difference between players is more than 500 points (the number is arbitrary), then the game should not be counted towards the rating!!!! This will encourage players to only play against strong opponents if they want to move up! And not like Dubrovin Pavel - to play against those who have only 1-2 games or a rating below zero, and then complain that the reports were not confirmed :D

Let's argue now :)

I cannot agree with this message because:

firstly, the concept of a "strong opponent" is often subjective, and there may be quite a few strong opponents for a particular player;

secondly, playing against a "strong opponent" and playing against a "less strong opponent" gives me the same pleasure, and if such a restriction is introduced, as you suggest, then Heroes players will simply not find each other online, there will be fewer games, and any game brings some experience;

thirdly, I specifically do not choose opponents who are obviously "less strong". Sometimes I just don't want to get bogged down in complex calculations, but I just want to play for fun, so I don't rush into a fight with those I consider to be higher level, and when I want to play with them and I feel ready, then we play; And I don't consider myself a player of such a level that I should only play with the "best". Please don't judge me harshly;:smile10:

fourthly, I was talking about confirming the result not because my game is not being confirmed, but the game of my comrade, because in general I think it is wrong when, because the loser does not respond, the winner is not awarded the deserved points.

Well, I'm off to work, so I'll keep it brief. 1. Which opponent to play against, and how to protect points – so a weaker player will never be able to play against a stronger one. All games will be protected. 2. The rating has been in place for a while, and of course, it was intended to be refined. I'm actually surprised that it turned out to be quite successful, working without problems for half a year. 3. A situation has arisen – the one with more wins is ranked higher. Well, it's inevitable, there's no getting around it. 4. Points for Tournaments – maybe someone has forgotten, but this proposal was already made by someone, and in the discussion, they decided not to introduce it, probably for now. And now this question has arisen. 5. Points for Tournaments should be introduced. I agree. But, of course, it's also necessary to take into account: - the representation of strong players in the tournament; this is difficult, and it's not easy to derive a pattern. Using the rating might not be entirely correct, although on the other hand, in a sporting context, this can be seen as an analogy to a decline in form for an athlete (let's not laugh, this is just an example). That's why he's ranked lower than in the previous season, even though he's a strong player. - the number of players - of course, online play is taken into account. More specifically: Regarding points – approximately from memory – it turns out that one player usually wins no more than 1 tournament per season (correct me if I'm wrong), so we can safely give 500 points for a victory (if he doesn't play that well, as practice shows, 500 points still won't be enough, but it's right to encourage this). But then the question is – should his matches in that tournament be taken into account or not, because points are also awarded for them; this needs to be discussed. And in general: 1-500 2-400 3-300 Why not split it in half with the lower-ranked player – a 200-point difference for winning against 2 players (approximately – in elimination games) is normal, + probably the victory will also be calculated in this way (as a game), so I think it's okay. These numbers are also supported by others, so I just confirmed them. And if we calculate it, taking into account the matches in the tournament + the bonus for winning, you can score about 900 points in a tournament (this is the minimum), with the bonus, it doesn't seem like much. Okay, I'll talk to you later. :smile17:

Regarding points, I suggest the following formula: Where R is the rating for the Tournaments, P is the participant's place in the Tournaments, and N is the number of participants in the Tournaments! And you can reward everyone! The last one will receive little, but still something... Or you can only reward a certain number. For example, like this:

1-4 participants ----- 1st place
5-8 participants ---- 1-3rd places
9-16 participants --- 1-5th places
17-32 participants --- 1-8th places
33+ participants ---- 1-10th places

2Dubrovin Pavel: Let's stick to our own opinions

2 Derty
I agree that it would be better to develop a formula based on which the calculations will be carried out, but your proposed formula gives too many points. For winning a Tournament with 64 players, 1200 points are awarded. IMHO, it needs to be reduced a bit, for example: R=150/P*sqrt(N)/2 or something like that. Regarding additional points for 1st-10th place, hmm, I don't object. Not all players can claim a place in the top three, and this will encourage players of all levels to participate in Tournaments, because almost any player can take 10th place.
Guys, let's be bolder and more active! Otherwise, it will start again: 5 people discuss, and the rest watch, but when the changes come into force, complaints will begin, like no one asked for their opinion. We also need to encourage participation in inter-forum Tournaments! The Portal players were rewarded for winning against us, and quite well by their standards. I understand that if they had lost, they would have also received some bonuses. And we just thanked each other! Judging by the upcoming changes on our forum, each section will have its own coordinator. We can arrange it so that points for such inter-forum Tournaments are awarded not according to the standard formula (which we will come up with and approve ourselves), but more, in agreement with the relevant coordinator.
I ask everyone to participate more actively in this discussion. IMHO, this is very important, and there is not much time left! Please express your opinions with specific proposals and figures!

2invisible:
What do you think, how often will tournaments with 64 participants be held? I think 1200 points for a tournament with 64 participants is a good amount, considering that second place will receive 600, and third place only 400... not that much, but the person who wins such a tournament will be mega-cool and will definitely deserve to immediately rise to the top of the leaderboard!

Although, maybe I've slightly overdone the formula... here are three options for changing it:
1. Replace the square root with a cube root!
2. Replace N with N/2
3. Replace 150 with some other constant.
4. Leave it as is :)

Well, I've generally proposed the structure of the formula, and let other people who monitor changes in the rating and know what prizes are needed for whom refine it!

Separate prizes should be established for inter-portal battles! The residents of the Necropolis were rewarded very well! We should take them as an example.

2 Dirty Although you might be right, because if a tournament for 64 players is held, it will be at most once per season. If the winner is awarded 1200 points, it's not critical, because we can easily earn much more points in the rating :) And if a winner is found who has outplayed 64 players, then they rightfully deserve to be in 1st place in the ranking!

I'm in favor of adding a bonus to the rating for winning a tournament! The tournament could be given a weighting factor, depending on the average ratings of the players participating in it. The bonus to the rating for winning, or finishing in the top three to five places, would depend on this factor.
And yet, we still need to come up with some kind of productivity indicator that reflects both the rating and the number of games played. Kostyan, your example of an exceptional situation, in my opinion, will never occur in practice! Just as with the fictitious tournaments that Yegor is talking about. No one prevents you from creating a few clones in a season and boosting your rating with them. However, this doesn't happen, does it? Or does it?

No one prevents players from creating a few clones during the season and using them to boost their rating.

You can't really boost your rating significantly with clones. Perhaps some do, but:
- in my opinion, it's not worth it.
- the creator of the clones understands the risks involved. In case of exposure, they face a lifetime ban. Want to know how this is detected? It's very simple ;) All it takes is the desire and a reason to verify the authenticity of all players to check for the presence of clones.

I would like to suggest the following:

1 - 8 (3 rounds / 5 games - winner takes all)
9 - 16 (4 rounds / 6 games)
17 - 32 (5 rounds / 9 games)

_________I__________II__________III
1-8------400---------250---------150
9-16-----555---------350---------250
17-32----777---------450---------300

-----------------------------------------------

And in seasonal games, deduct 75% of the winner's points for a loss, instead of 50% of the points earned in the game!!!

2scorp: The idea is good, but you need to reward not only the top three places. In a tournament with 32 people, you should reward at least 5 places, maybe more!

Regarding points, I suggest the following formula:

Where R is the rating for the tournament, P is the participant's place in the tournament, and N is the number of participants in the tournament! And you can reward everyone! The last one will receive little, but still something... Or you can only reward a certain number. For example, like this:

IMHO. It’s okay.

Although I've been here for over five months, I still feel like a newbie, I haven't even had time to read the entire forum. Nevertheless, I would like to express some thoughts. A rating is an individual numerical indicator of an athlete's skill. I would like those who look at our rating to get an objective idea of the players' skill level. What we have now is not a rating, as it generally does not reflect the level of play. The system of making adjustments based on tournament results, which is currently being discussed, will certainly add some objectivity, but it will not solve the main problem. As far as I understand, some other rating system is being prepared. Eugene, please tell us more about it. A system that objectively reflects the class of players is very much needed. This would simplify the work of tournament organizers, both when forming a team (in tournaments like the inter-portal one), and it would also be possible to "seed" players in internal tournaments. The table we have now very well reflects the statistics of the games and, in my opinion, is a very valuable product for careful study. When I first registered on HW and started playing, it was important for me to know what level of player I was fighting against. Therefore, I developed my own "rating" based on this. In fact, this idea periodically appears in forum posts. Players who lost all their games and those who only won against them were "removed" (this also included games like vitalic def Neoandy - for the latter, this was his only game of the season). Then, a table of game results was compiled for approximately 20-30 players. Again, players who lost most of their games were excluded. Then, the percentage of won matches between the players at the "top of the iceberg" was calculated. The result of my theoretical research was then well confirmed by an experiment - a player vote conducted by Nagan. (I may have made a mistake somewhere, don't be angry.) Winter season (games/wins/%): vitalic – 12/10/83, Mizrael – 3/3/100, Yakudza – 8/7/88, uran – 5/4/80, Dreamcatcher – 11/7/64, Alex_R – 4/2/50 invisible 32/15/47 Captain 9/6/67 ewgeniy 4/2/50 Деконтаминатор 34/10/29 vbn 15/6/40 Vigo 10/4/40 Дубровин Павел 14/3/21 Kamikaze 3/1/33 Unfortunately, in the winter season, many players had too few games to be sure of the accuracy of the statistics. But the spring season provided more material for reflection. Spring season: vitalic 25/21/84 uran 14/10/71 scorp 39/23/59 Деконтаминатор 47/26/55 invisible 48/25/52 Mizrael 23/12/52 Rhox 20/10/50 vbn 21/10/48 Dudge 11/5/45 Pit 27/12/44 Дубровин Павел 61/23/38 Javdet 28/10/36 Outlaw 25/8/32 Volland 10/3/30 Dreamcatcher 20/5/25 Aldegir 39/9/23 RetroFM 24/5/21 Over a year and a half of online games on HW, a large amount of statistics has been collected. We could think about how to improve my system. And based on it, form a "top" and "first" league, as is customary in other sports. In parallel with the seasonal games, a seasonal championship could be held. For example, 8 players from the "top" league play a round-robin tournament - 7 games in 3 months. The seasonal champion and 2-3 outsiders are determined. The same applies to the "first" league. In the next season, 2 players from the first league move up... Based on the seasonal games, newcomers and those who have improved their game level get into the first league. Otherwise, we end up with Federer being awarded points in the rating for the points he scored in junior school.

Huge thanks to Pit - for the sake of LM!!!
His tables reflect the actual strength of the players much better than the CC-D table. And therefore, it exposes its shortcomings... The rating should be changed, in my opinion...
I also support the idea of Leagues.

Many thanks to Pit - for the sake of LM!!!
His tables reflect the real strength of the players much better than the CC-D table. And therefore, it exposes its shortcomings... The rating, in my opinion, needs to be changed...
I also support the idea of Leagues.

I agree, everything is correct, but now let's talk about reality.
1. Discussions are just discussions; everyone understands that nothing is perfect, but no solutions have been proposed yet.
2. The percentage ratio is correct, but there are many nuances in this - it's a very complex calculation (if you understand what I mean). It's not just about counting who won more games against the top ten - but how do you determine this top ten, how do you link the percentage and the place that doesn't exist yet - because the percentage will change depending on this game, and so on. It's difficult to think through. Whoever comes up with it will be great, but of course, they will again try to make it simpler, so it's easier, and again, it won't be perfect, plus don't forget that you also need to write a program for all this.
No one will calculate this on a piece of paper; have you tried? (I have).
3. Here's a solution regarding the rating:
- what exists now will be like statistics, and at the end of the season, when all the games are played, you can calculate the percentage and determine the winner. It's a fresh idea.:smile44:

And what comes out of all this discussion is that it's no longer profitable to play against weak players, and they may simply not want to because of the rules, which is a minus. But if you want to play against a strong player...
Regarding the league, it's already like a tournament.
Dividing players into those who are not immediately competing for victory - beginners - is not correct (for example, Scorp - he only started playing quantitatively in the season when he won, and he practically didn't play in the winter, and if there were rules like the proposed one, he wouldn't have won, there wouldn't have been access to compete for 1st place), and in general, there are many nuances here.
Let's discuss and discuss.