Posts from Player rating for tournaments across all installments of Heroes.

2AlexeyPank

4. Ultimately, a bonus that takes into account the strength of the tournament is necessary. And so far, no one has put forward arguments against the refined version of it. I would like to hear constructive criticism.

Your bonus has one serious problem – its scale. Its idea is taken from tennis, where, if I'm not mistaken, it serves as a nice addition for a player who has gone through a more challenging bracket. The main points are determined by the rank of the tournament and the round in which the athlete reached (i.e., their placement). Major tournaments are assigned a rank in advance, and players know what they are playing for, and the prize money is accordingly high. In our case, we cannot know the rank of the tournament in advance, because there is a strong element of randomness in the participants. That is why I proposed introducing a coefficient for tournaments, which was actively discussed with AmberSoler. Yes, there were shortcomings, but still, the coefficient would quite well indicate the strength of the tournament and differentiate it accordingly. The difference in the final rating of the winners of different tournaments could be several times greater. It all depended on the composition. But your bonus so far does not exceed 10 points. It is completely unsuitable for separating tournaments by strength. But as a stimulus to progress – yes, that's true. But even in this version, it should be adjusted: negative bonuses should not be awarded (being responsible for one's own previous success is too much), and the winner should still have the maximum rating. But is it worth the effort?..

This just means that you will choose the system in which your personal result will look better.

I won't assert or deny anything (it's pointless), but what I wrote has a different meaning: I will choose after carefully considering and within a reasonable timeframe, and what you understood is not interesting to me... and I will try to keep my rating in the same place across all systems. :)

The three options that were already in use were proposed BEFORE the start of Kanighem's tournament. And no one knew how it would turn out, including the developers of these systems.

The rules for participating in tournaments did not include a clause requiring the selection of a scoring system... the complaints are not directed at me!!!
P.S. In the future, comment only on what is written... and don't try to guess what I mean!!!

But is it worth the effort?

I think it's not worth the effort. All possible calculation options that do not take into account the position (or similar parameters) turn out to be quite cumbersome, difficult to understand intuitively, and inconvenient to use. In essence, Iron's option is optimal in terms of simplicity, clarity, and relative accuracy. If we assume that the average tournament does not have large gaps between the leader, then, given the rejection of the position, this method can be left as a working one... No "simple" modifications can be applied to this calculation option, because we immediately move towards the same problems that we want to avoid. But when there is a gap between the leader, there is a bias towards an unjustified lowering of the rating of all participants except for the first place. Moreover, the players who are higher in the tournament table suffer the most. A simple example: Day 1 - 100 points Day 10 - 10 points Day 100 - 1 point. The gap between the first and second place is equal to the gap between the second and third place (ten times). But in this case, an unreasonably small number of rating points is awarded for second place... From the point of view of the total rating, the results of the second and third players are practically equalized, which is not actually the case... Moreover, regardless of the number of players in the tournament, all other results will be between 1 and 10 points. And the reason is chance, namely the appearance of a strong player in the tournament. If he had not played, the result would have been different: Day 10 - 100 points Day 100 - 10 points. As you can see, there are no significant changes for the outsider, while a relatively strong player is highly dependent on a random combination. Moreover, this effect always exists, it just manifests itself to a lesser extent in the absence of a gap in the results. If we ignore this point, we can stop the discussion about the rating until better times.

I think it's not worth overcomplicating things.

No, no, we still need to talk a bit more :)

All possible calculation options that do not take place (or similar parameters) into account turn out to be quite cumbersome, difficult to understand intuitively, and inconvenient to work with. Essentially, Iron's option is the optimum in terms of simplicity, clarity, and relative accuracy. If we assume that an average tournament does not have huge gaps created by the leader, then provided we abandon the place ranking, this method can remain as a working one...

Well, finally it's been put into proper professional words—exactly what I was trying to explain so clumsily :)

No "simple" modifications can be attached to this calculation option, because we immediately drift back toward the same problems we want to get rid of.

But we were saying that it could be slightly improved if something was introduced into the formula so that a VERY small bonus could be given for secondary parameters, even just to separate players with identical results in the primary parameter within one tournament. You promised you'd think about this question, didn't you??

But when a leader pulls ahead, a tilt occurs toward an unjustified lowering of the rating for all participants except first place. Moreover, those players who are higher in the tournament table suffer the most.

That's not quite true!!! Or rather, it's completely untrue!!! You write that if there is a strong gap between the leader and the bulk of the players, what does that mean? It can only mean one thing: that in this particular tournament for this version, there were no more high-tier or mid-tier players. Consequently, it's not an unjustified lowering of all participants' ratings, but rather an UNJUSTIFIED INCREASE in their rating is being ELIMINATED! At least relative to other players of the same level for this version who DID NOT PARTICIPATE in this tournament. This is what needs to be understood in order to abandon taking place into account when calculating the off-rating.

A simple example:

1 day - 100 points
10 days - 10 points
100 days - 1 point.

The gap between first and second place is equal to the gap between second and third (tenfold). But in this case, an unjustifiably small amount of rating points is awarded for second place... From the perspective of total rating, the results of the second and third players are practically leveled, which is not actually the case... Moreover, regardless of the number of players in the tournament, all other results will fall between 1 and 10 points. And the reason is chance—specifically, the appearance of a strong player at the tournament.

It's not about the appearance of a strong player, but the absence of other strong and mid-tier players. In your example, the gap between second place and the leader is 10 times, but in real numerous tournaments where players of all tiers are represented (strong, medium, weak), we see how close the results are for the primary parameter; however, there are major moments when many strong and mid-tier players for a given version fail to show up (example with stage 4). And there, the gap between first and second place was only 3.5 times, contrary to your example where it's 10 times. But it's possible that the player in second place belongs to the same mid-tier group of 4, but he was unlucky and made a mistake (didn't use space or graves, even though it was allowed by the rules, and consequently fell far behind the leader in days). Should a bonus be given for that? After all, he himself said that toward the end of the game he understood and saw how to get a result of 6-7 days. Now, if he had achieved that, then his rating under my simple system would have been normal, which numerically would be about 50 points. But since the second place result was submitted as 11 days, the rating is 27. What conclusion can be drawn? Either a weak player in this version took second place due to the absence of strong players, OR he is also a strong player but made a gross error on this map (just unlucky). In these cases, NO bonus should be given. But with the oscillator, it turned out that a bonus was given in these cases, and a decent one too!!!

If he hadn't played, the result would be different:

10 days - 100 points
100 days - 10 points.

As we see, for an underdog, no significant changes occur; at the same time, a relatively strong player is highly dependent on a random combination. Moreover, this effect always exists, simply

WHAT RANDOM COMBINATION DOES A STRONG PLAYER DEPEND ON???!!!
I don't understand how you are comparing them. One strong player showed a result of 1 day (see your first example), and another strong player showed a result of 10 days ON THE SAME MAP from the same starting save. How can you call him a strong player (or say he was very unlucky)? Theoretically, there should be results of 2,3,4,5,6,7,8,9 between 1 and 10. That's why the question arises regarding example 1. Why do you assume that a player with a result of 10 is strong? Well, sure, if he had a good rating and showed good results for this version BEFORE, then yes. But then the simple explanation is—he had little real time and didn't play the map to his full potential or MADE A MISTAKE. I repeat, NO bonus is DUE for that! even though he took second place—it just happened spontaneously in this tournament due to the chance absence of players of a certain level, and including the same place in the formula will only unjustifiably boost such a player's rating.

it manifests to a lesser extent in the absence of a gap in results.

Everything is fine with gaps in results. Everyone gets what they played for. But here's the snag: if the leader is DEFINITELY a weak player (see your example 2), that remains in my system too (this flaw exists in yours as well). All players of this tournament will receive an actually inflated rating due to the complete absence of mid and high-tier players. But let everyone ask themselves—what is the probability of such a situation in official tournaments, even on our forum??

If we ignore this point, we can stop talking about ratings until better times.

That's clear, and let AlexPank calculate using all three systems if he wants, but what concerns me most right now is the question—by which system will the nominees for all versions and the universal be determined? And how will it be done?
I suggest again: let's aggregate results for the universal only for those players who played 5 or more stages. Moreover, each player should choose their best results from only five stages. The places in the universal nomination will be determined by the sum of ratings from one of the systems. Why is everyone silent about this proposal—is it bad again?

Added after 9 minutes

well, since you've looked at the table, please take the trouble to find the thread with the tournament regulations; everything is clearly explained there in the first post. And get something soft to lie on, because I fit all of this into 1 POST on 1 PAGE, and I'm not Yandex... I gave one link and that's enough, otherwise you'll actually get spoiled!

What do I have to be spoiled for—I looked around the online section and understood and found nothing. What am I supposed to do—plow through every single post? I believe that if there is a rating calculation system, its description should be in a visible place, as well as the Player Rating table. Didn't find it. :smile19: Removed :smile19:
Kyle: no one is to blame that you cannot find the right one among three threads! This is entirely your problem. And "attacking" other people over this is simply not nice!
Respectfully.


Added after 29 minutes
PS. And I also forgot to remind you about the co-ops. They can actually artificially create the presence of players of a certain level at a tournament, even if not all of them have that skill level. And thereby, when taking place into account, the rating of those they push down will be unjustifiably lowered. Some might object that those players will still receive an unjustifiably higher rating anyway—well, to hell with them; the main thing is that the rating of those following them in a system without places WILL NOT DECREASE relative to their real result, and that's what matters. The same applies to the presence of a cheater at the tournament. In a system with place ranking, he will push honest players down and LOWER their rating. In a system without place ranking, he might get a better rating if he isn't caught, but the ratings of those following him won't suffer due to an unjustified lowering of their rank.

No, no, we still need to talk a little bit :)

Why should I mess around? I browsed the online forum and didn't understand or find anything there. Am I supposed to go through all the posts? I think that since there is a rating calculation system, its description should be in a prominent place, and there should also be a table with player ratings. I didn't find it. :smile19: Removed :smile19:

Actually, yes!!!! :smile19: Removed :smile19:
Let's close the topic about the online mode, at least regarding the 3rd game in this discussion, as soon as one of the rating calculators can find the relevant topic, and even more so, understand the specifics of the online mode. I will definitely delve into the rating formula and I guarantee 99% that I will adjust everything according to your wishes. In the meantime, let's each do our own thing and not argue or be sarcastic.

Kyle: yes, let's not be sarcastic. This applies to you too.
Sincerely.

:smile19: Deleted :smile19:

I will look into the rating formula and I guarantee 99% that I will adjust everything according to your wishes. For now, let's each focus on our own tasks and avoid arguing and being sarcastic.

:) You have already been told that you don't need to recalculate anything according to our wishes. The ratings for the offline and online versions cannot be calculated using the same formulas. I just want to learn about your calculation system; please provide a link to the topic where it is described.

Kyle: One more message like this, and I will start taking appropriate measures!
Sincerely.

/topic/at/77165/
5 и 6 пункт

Gentlemen,
I am shocked! This kind of behavior is simply unacceptable.
If I see any more insults or veiled "attacks" on opponents in the thread, measures will be taken.
Let's respect each other.

Sincerely.

/topic/at/77165/
Points 5 and 6

Thank you for the links. Well, I couldn't have known that everything related to this system could be summarized in two points and is located in the Tournaments rules. I was looking for a separate thread about discussing and analyzing the rating calculation method – so naturally, I couldn’t find it. Why would I reread all the posts in the threads? That's why I asked for links. If you are interested in my opinion on the system, tell me where to write it; apparently, it shouldn't be here. Or should it be here?

If you're interested in my opinion on the system, let me know where to write it. It seems like it shouldn't be here. Or should it be here?

You can write it here, as this is the appropriate topic. ;)

You can post here, as the topic is relevant. ;)

Maybe I'm wrong, but what's written in points 5 and 6 isn't a rating system, or I'm not understanding something. It's a knockout tournament, and points are used to determine the semi-finalists and finalists: 3 points for a win, 0 for a loss, and 1 for a draw. These are somehow called "rating points." This is like a football championship where a certain number of points are awarded for a win, loss, or draw. But can this be called a rating? These points are intended to ultimately determine the TEAMS' PLACES in the championship. After the places have been determined, you can calculate the rating. In principle, if there was only one championship in Adelaide, then perhaps such a system where the points used to determine the places would also serve as rating points would be suitable. But she will have at least two championships a year. And two more next year. How will the rating be calculated – by summing these points? Let's take two championships for simplicity. I won't bother you with numerical examples for now; maybe it's clear enough as it is. In my opinion, it's possible that after two championships, two players will have the same total number of points, but their places will be different. Is this possible? But with different places in the championship, it doesn't make sense that their rating should be the same. In this case, you can apply the AmberSole formula, where the rating is calculated based on the place taken (this system is very suitable for online play). And in this way, the points for determining the semi-finalists and finalists will not be confused with the Rating Points.

Hmm.... :smile19: Deleted :smile19:
In general, you misunderstood two points, and it's not required of you, but your understanding of this issue and your suggestions have been read, understood by me, and will not be taken into account.
We also have other online tournament organizers; you can try to convince them that everything they do is wrong and offer them some of your own or Soler's formulas. Perhaps my mistake was that I even wrote something in this thread; ignore it, and that's all.

Hmm.... :smile19: Deleted :smile19:
In general, you misunderstood 2 points, and it's not even required of you, but your vision of this issue and your suggestions have been read, understood by me, and will not be taken into account.

And congratulations on that :) Did you think I doubted it? Not at all :)

AmberSoler-y
I also want to draw AmberSoler's attention to an incident at the online tournament. I saw how many clones were created. And the main thing is, I didn't understand what it was for. This guy ruined the tournament for all the players. And in the case of an offline tournament, he would have also messed up the ratings based on the oscillator, if he were smart enough and achieved good results. I'm saying this because there are idiots everywhere, and no one knows when they will appear in our offline tournaments. They usually appear when you least expect it. Therefore, it is better not to take the place in the offline rating system into account. Here is another real-life example as proof.

Three questions: does anyone know when the voting will end?
Or will the discussion continue indefinitely? Perhaps we should set strict deadlines?
Best regards, Draco

Well, we'll definitely revisit this topic until we find the perfect formula... which means it will be ongoing!
Regarding the King's Bounty tournament, it would be ideal to have everything finalized by the beginning of the last stage, which will be in November (in 3 months).