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

Frankly, I'm tired of discussing this – everything has already been written and rewritten here. I believe that my system, specifically, shouldn't be used for calculating Grail points. There's no need for any aging mechanism in the Grail points calculation. The oscillator proposed by AmberSolero would be suitable. That is, the non-linear part is calculated as I suggested, and the linear part is calculated using AmberSolero's formula, taking into account the significance of the occupied space. Then, we take the arithmetic mean and award the player Grail points. I think you can't come up with a better system for awarding Grail points.

That's what I was talking about – AmberSolero's system (for calculating the GRAIL RATING!), I don't understand why you didn't understand:o. I'm not going to abandon your system at all, and we will continue to calculate the GAME RATING using it as before.

Sincerely, VDV_forever

I want to clarify the bonus points issue. I don't think I was understood correctly. Here's what I meant. After each tournament, a ranking grid is created from the submitted save files, based on their previous ranking. And the player who takes first place in this grid is considered the strongest player and should be in first place. It doesn't matter what their overall ranking is; what matters is their position in the specific grid. Thus, if a player is in 100th place in the overall ranking but takes first place in a specific tournament grid, their bonus will be calculated based on their first-place finish. And even though they took first place, they won't get a significant increase because the others are even weaker. But if they participate in a tournament where they are 25th in the grid, and they take first place (agree, this would be a breakthrough - beating 24 stronger opponents), they will receive a substantial bonus. Only if you play at or above your level in a tournament will the bonus be positive. I am more inclined towards the formula B = N / (R(i) - M(i)). The values will be small here. But if we want to reward those who make a significant breakthrough, it makes sense to use a different formula - B = N / (M(i) / R(i)). Where N is the number of players, R(i) is the player's ranking, and M(i) is the player's position in the tournament. This way, there shouldn't be any such issues.

I'm ready to calculate the rating for the second-tier heroes (three tournaments have been held).
If needed, please provide the calculation methodology; I don't see any clear formulas in the discussions that everyone would agree on.

I'm ready to calculate the rating for the second-tier heroes (three tournaments have been held).
If needed, send me the calculation methodology; I don't see any clear formulas in the discussions that everyone would agree on.

If I'm not mistaken, almost none of the Humans who participated in the first tournament are left now.
Although, okay, let them be. Maybe we can test the "aging" of the rating in practice.

I am ready to calculate the rating for second-tier heroes (three Tournaments have been held).
If needed, send me the calculation method; I don't see clear formulas that everyone would agree on after discussions.

It has already been calculated... check it out here! :smile01:

I am more inclined towards the formula B = N / (R(i) - M(i)). The values will be small here.

Typo. It should be the other way around - (M(i) - R(i)). And by R(i), you mean not the rating, but the position in the ranking, right? There is an inverse relationship here; the higher the rating, the smaller the number representing the player's position... :) Because the rating itself can be any number in the range of 0-1000.

But if we want to reward those who make a sharp breakthrough, then it makes sense to use a different formula - B = N / (M(i) / R(i)). Where N is the number of players, R(i) is the player's rating, and M(i) is the player's position in the tournament.

I think it would be more correct to write the formula as follows: B = N * R(i) / M(i). The bonus will be in the range of [1; N^2], which is very critical.

Example: 50 people participated in the tournament. The tournament winner - M(i) = 1 - is participating in tournaments for the first time, which means his current rating is 0, i.e., R(i) = 50. And what will he get as a bonus? B = 50 / (1/50) = 2500 rating points... :) What is the initial mistake? It is that you assume that the current rating indicates the player's strength. But this is not the case. And we have been talking about this for a long time - the current rating has nothing to do with the player's strength. It shows the player's activity, taking into account his strength, which, I agree, is not the same thing... Any newcomer, as well as any inactive player, has a zero rating. This does not mean that they play poorly, and therefore the bonus should not have such sharp jumps...

The method of calculating the bonus from the average rating of all tournament participants, which I proposed earlier, will not lead to such an imbalance because it takes into account all players at the same time, and not just one... Before this idea was voiced, it was, of course, repeatedly tested... :)

Therefore, I ask everyone who wants to demonstrate their calculation method to first test it for extreme situations, see how the function will behave, and assess the probability of these extreme situations occurring... The example above shows that in an extreme situation, a player will receive an unattainable bonus, and the probability of this situation occurring is 100% :) Because the tournament winner may very well be participating in a tournament for the first time... :) Also, a strong player can easily skip a few tournaments on purpose to lower his current rating in order to then increase his rating after a brilliant performance. Rating calculation is, first of all, logic, and then mathematics.

I was talking about the AmberaSolera system (for calculating the GRAIL RATING!), I don't understand why you didn't understand:o. I'm not going to abandon your system, and we will continue to calculate the GAME RATING using it as before.
Best regards, VDV_forever

Kyle correctly pointed out my mistake. Maybe we don't need the oscillator, but just take the linear part of the formula, that is, what AmberSoler added. Then, the last place would simply have 0 grails - that's unfair. But there is a solution. A threshold for the underdogs. That is, we calculate everything using the linear formula, but everyone must receive at least, say, 10 grails, even if the formula gives 0?
This linear part is a normal formula. It takes into account both the number of participants and the places taken.

Added 43 minutes ago

What is the initial mistake? You assume that the current rating indicates the player's strength. But that's not the case. And we've been talking about this for a long time - the current rating has nothing to do with the player's strength. It shows the player's activity, taking into account their strength, which, I think, is not the same thing... Any beginner, as well as any inactive player, has a zero rating. This does not mean that they play poorly, and therefore the bonus should not have such sharp changes...

The current rating is related to the player's strength, for God's sake, who needs to take into account only activity? And I would say the opposite - the current rating shows the player's strength, taking into account their activity. But here you need to understand that a new person with 0 is a dark horse if we haven't seen him anywhere. It's different if a strong player has fallen in the rankings due to missed games - this table won't tell an outside person anything, but interested players and the coordinator know that the player just didn't play. And the current rating shows his activity at the moment. And he will show his strength when the player comes and scores 100 rating points, which will sharply raise him in the rating table from last place to the middle. But with a 10% aging, this will not be so noticeable, well, okay.
The current rating is maintained for those in the know who follow the events. It is especially important to think about publishing not only the latest figures of the current rating for everyone, but on another page, in Excel format, to keep track of its dynamic changes for everyone, so that it is possible to distinguish where activity is shown and where strength is shown at a given point in time for a given player.

Added 1 hour 38 minutes ago

Are you kidding? If you support the oscillator with the same energy with which you disputed it, then there is nothing for me to do there... Feel free to use the formulas and calculate them. :) It's all sophistry... Grails are not an indicator of anything.

Grails are a reward for prize-winning places in tournaments. Well, if not the oscillator, then I would support the linear part with a threshold for the underdogs. VDV decides how to spend the grails.

Added 34 minutes ago
First, logic, then mathematics. Or what I was thinking about.

The current rating is an indicator of the player's strength in Heroes at the end of the current tournament, taking into account the results of several recent tournaments, the number of which depends on the size of the aging of the rating you have before the start of the tournament (the so-called old rating). The current rating of a player takes into account both the strength of active players who participated in the last tournament ON THIS RESOURCE, and the activity of passive players who missed the tournament ON THIS RESOURCE; the rating of the latter is simply reduced by 10%, while active players try to compensate for the aging of the rating with their tournament rating in order to increase their current rating.
Why suddenly reduce the rating of those who missed it? Maybe they were playing in other tournaments on other resources and were no less strong?
1. There is no general rating system between resources. Therefore, as a coordinator of the Current Rating for VDV's Heroes 3 tournaments, I don't care who, where, and how cool plays. The rating is also not interested in knowing why the player missed the tournament on this resource. The penalty for missing a tournament is 10% of the old rating. If this is still unclear to someone, it's not my fault.
2. I can correct the summary table, but I will not reduce the rating in only one case: - at the personal request of VDV. If you personally prove to him that you missed this tournament because you participated in another tournament on this resource, even if it is a tournament not in Heroes 3, or on another resource (for example, you defended the honor of HW on the road). To do this, you probably need to provide VDV with a link in a private message to the results of the tournament where you played. Since my rating is experimental, it is even possible to calculate the rating for a given player in individual cases (for now, this applies only to Heroes 3) and add his result to the summary table of the Current Rating on this resource. To do this, I need to know only three parameters from that tournament - the type and size of the map on which the tournament was held, the result of the leader of that tournament in days, and your result in days. THAT'S IT! This is the first step towards a general rating between resources. If you start trying to take into account the Place Taken and the Number of Players, everything will become very confusing. Given that there is also confusion with ratings on other resources, I do not want to include the coefficient of the strength of the game in the formula.
Otherwise, how will I calculate the rating for a player from our resource who fought in a tournament on the road and does not want to have a 10% reduction for no reason? And again, I repeat - I just need to know three parameters.
Especially if we are going to promote this rating to the status of a General one. I had to consider several examples to understand this. We didn't want to find out the reason for missing players, but it seems that we will have to give players a chance to keep their rating if they wish. I think this idea should be pushed further.
3. If you don't participate in any tournament for several times in a row, and your Current rating is falling, but you still consider yourself a strong player - that's your right. But other players participating in tournaments don't care, and the Current Rating on this resource doesn't care either. How can the Current Rating know - do you still remain a strong player or are you slowly degrading? Come back to the tournaments and prove your strength. The current rating will react to this immediately. Or promptly provide VDV with links to your results on other resources in the glory of HW!
4. In speed tournaments, I will use the permissible limits for completing the map to determine the leader's coefficient. If the tournament is for experience or resources, then I will have to calculate without permissible limits.

I realized my mistake. I was thinking about it yesterday. What if the formula is: B = ((R(i) - M(i))/N) * sqrt(K), where N is the number of participants, R(i) is the player's rank in the rating of this tournament, taking into account the ratings of all players, M(i) is the player's rank in the tournament, and K is the tournament coefficient, equal to the average rating of this tournament for all players. In this case, there may be negative values if the player performed significantly worse than expected.
I'll try to recalculate the tournament results for Heroes of Might and Magic II using this formula and see what happens.

Here are the results, including my bonus. I've attached them. The changes are minor, but they are there.

Here, I calculated it with my bonus. The results are attached. The changes are minor, but they are there.

So, what's next? Here are the calculations for the same tournaments using three different systems. And yet, no one is commenting. Yes, there are distortions everywhere, but which one best reflects the players' results? My comment is the first one. Even though you call my system non-linear, in this SPECIFIC example with two tournaments, the player rankings look more linear than in other systems. And I know why. The dependence on the position taken introduces distortions in the linear part, and AlexPanka's bonus introduces additional distortions. OlegRS 17 48.5 59.8 63.8 19.6 24.2 37.3 41.9 19.6 20.4 40 Analyze this player's line. With my system, he has a rating of 63.8 for the sum of the two tournaments, 41.9 with Amber's system, and 40 with Alexy's system. Your statement about minor changes is understandable. But what can be said about them when we see significant changes introduced by the linear part? What a difference, almost 22 points less, due to the fact that the player happened to take last place in the second tournament. Compare Kyle (9th place, 68 days) and Nekto (10th place, 69 days). In the second tournament, neither of them played. The final rating with Amber's system is 10.3 versus 5.2. With my system, it's 10.6 versus 10.4. The distortions in the middle part of the table are less noticeable, but that doesn't mean they don't exist. If AlexPanka's system is applied only to the linear part, then it's still somewhat acceptable. I told AmberSoler about this, that the dependence on the position can be added to my system, but in such a way that this parameter does not have a dominant influence over the "Days Played" parameter. AlexPanka was interested in this, and he started working in this direction. I think he succeeded. But the calculations are more complex, and it's very difficult to check. And the changes are minor, as AlexPanka correctly says. Now, try to ask Kyle to recalculate the ratings for all clans using your additional coefficient. Where he will have to process a lot of tournaments - let's see what he says afterwards - whether this coefficient based on position is needed or not

But the calculations are more complex, and it's very difficult to verify... We'll see what he says after he has to process a lot of tournaments – whether this coefficient based on placement is actually needed.

This is difficult for you, as you've repeatedly said. Those who understand mathematics won't see any complexity in the calculations. And you probably just made a joke about "a lot of tournaments," right? I don't want to go into the rest because you have a very narrow view of things, unfortunately... You didn't even understand what my argument in the previous post was about:

What is the initial mistake? It's that you assume the current rating represents a player's skill. But that's not true. And we've been talking about this for a long time – the current rating has nothing to do with a player's skill.

The argument was against using the current rating of a specific player as a criterion for calculating the bonus. The reason is as above. It's all explained in detail with an example. And you've latched onto your rating again... You're ready to "punish" a player for missing a tournament (which is absurd), but you can't accept the fact of being "punished" for last place. :) And the second one is purely a game mechanic, unlike the first... I didn't even touch on your calculations... But you keep seeing accusations that your theory is flawed... It's not even interesting anymore. Read my post again, and maybe you'll understand what I was trying to illustrate with the example... Honestly, I'm having a "déjà vu" moment. I stopped trying to have a conversation with you a long time ago because all your comments on any topic end up practically the same way. I don't know how to put it more simply... You're not capable of seeing the obvious flaws in your system, and I can't point them out to you because there's no argument in the world that would convince you – you've decided that for yourself. What else is there to say? The ability to critically evaluate your work is a big plus. You don't have it. In history, there was a character – the Roman consul Marcus Porcius Cato the Elder, who became famous only for ending every speech he gave in the Senate on any topic with the words "Furthermore, I believe that Carthage must be destroyed" (Ceterum censeo Carthaginem esse delendam!). And he didn't care whether the Senate had gathered to discuss celebrating the upcoming New Year or, say, to pass a law on elections... In my opinion, you are his follower... ;)

This is difficult for you, as you have repeatedly said. Those who understand mathematics will see no difficulty in the calculations whatsoever. As for the "heaps of tournaments," you were probably joking, right?

Apparently, you didn't understand what I was talking about. I was saying that my system can be used more flexibly. How will you calculate a rating for a player visiting another league if no rating is kept there? In any case, using AlexPanka's coefficient there is problematic. We already know why using the place and the number of participants in the calculation in the form you proposed is undesirable.

I have no desire to delve into the rest because your outlook on things is too narrow, unfortunately...

Perhaps. But with such a simple formula of three variables as the one I proposed, it has greater possibilities than formulas proposed by people with wide views and horizons. Perhaps the problem of rating calculation should have been looked at more simply. Which is exactly what VDV did in his system. Well, that one is far too simple and isn't linked to anything except Place. Note that although the voting in this thread may not reflect reality, these two systems, for some reason, are still slightly leading. People understand how their results will be evaluated through them, and if they wish, they can check for themselves whether they've been accidentally cheated :).

You didn't even understand the point of the argument I made in the previous post:

One must operate with terms more carefully, especially since everyone understands them in their own way.

The argument was against using a specific player's current rating as a criterion for calculating the bonus. The reason is — see above.

I agree with that. I disagreed that the current rating does not reflect a player's strength in Heroes.

Everything is laid out in great detail there using an example. And you've grabbed onto your rating again... You are ready to "punish" a player for skipping a tournament (absurd), while at the same time you cannot accept the fact of "punishment" for last place. :) And the second is a purely gaming moment, unlike the first...

I am not punishing the player. If he hasn't played, the rating cannot assess his current strength, so it begins to assess his activity. And punishment for last place, according to your own theory, is not good. Why did the person try? Well, they can't play well yet, so why further torment them by cutting their already smallest tournament rating in half using an oscillator or a small bonus coefficient :)

I haven't even touched your calculations... And you imagine an accusation of the insolvency of your theory everywhere... It's not even interesting anymore.

Then touch them. There is a calculation example for two 2-player tournaments. Kyle and Alex calculated and provided tables from real tournaments. I don't imagine accusations — they simply don't exist and cannot exist. And I am not accusing anyone, but showing why my system is better, and as long as the discussion continues, I will keep doing so — because you avoid examples and answers that are unfavorable to you and start attacking personally, which I do not do, unlike you.

Read my post once more and maybe you'll understand why I gave the example... To be honest, I'm already having "déjà vu." I gave up trying to have a conversation with you long ago, because all your comments on any topic end practically the same way.

Judging by this post, you haven't :)

I don't know how to put it more simply... You yourself are unable to see the obvious flaws in your system, and I won't be able to show them to you because there is no argument in the world that would convince you — you've

Show them not to me, but to others. When I write my posts, I don't expect only you to read them, because it's also useless to prove anything to you. You nowhere explicitly acknowledge the flaws in the calculation when considering place. At the same time, I have commented on the flaws in my system that you pointed out and have taken some into account.

decided for yourself. What else is there to say? The ability to critically evaluate one's own work is a big plus. You don't have it.

And do you? Come on, show me where you critically evaluated the non-linearity of the linear component??

There was one character in history — the Roman consul Marcus Porcius Cato the Elder, who became famous only for the fact that every one of his speeches in the senate on any topic ended with the words "And therefore I believe Carthage must be destroyed" (Ceterum censeo Carthaginem esse delendam!). And he didn't care if the senate had gathered to discuss the upcoming New Year's celebrations or, for example, to pass an election law... In my opinion, you are his follower... ;)

:) I liked that!
A person who won't read all 1000 messages :) exaggerating might drop by this thread. They will only read the last page, so I don't mind repeating the main conclusions regarding the systems under discussion. I wouldn't say that Marcus Porcius Cato the Elder was a fool :)
And therefore, since in your post I once again saw and heard a lot of new and interesting things about myself and the Roman consul, but did not hear an evaluation of the real rating calculations provided in the tables by Kyle and AlexPanka, then ....
"And therefore I believe Carthage must be destroyed" :D

:) I liked this! I wouldn't say that Marcus Porcius Cato the Elder was a fool... "And therefore I believe that Carthage must be destroyed" :D

Cato the Elder also wrote "Origins" - a work dedicated to describing the history of the Roman Empire, albeit in a rather peculiar way. All the facts that diminished the role of Roman virtues, for some reason, were not included in the book. With this, he laid the foundation for one of the modern professions, the name of which comes from his second name - Censor. The analogies are direct... :D

But essentially - I repeat. I don't consider the rating, so do as you see fit. A dozen tournaments will pass, and something might change there... If, of course,

Cato the Elder also wrote "Origins" - a work dedicated to describing the history of the Roman Empire, albeit in a rather peculiar way. All the facts that diminish the role of Roman virtues, for some reason, were not included in the book. This is how he laid the foundation for one of the modern professions, the name of which comes from his second name - Censor. The analogies are direct... :D

:) Well, I don't quite agree, although it might be essentially correct. But...
1. I agreed with the 10% aging.
2. I introduced acceptable parameters when playing for speed, as I acknowledged that there is also a flaw in my system due to the lack of strong players.
3. I acknowledged that I cannot even approximately indicate acceptable parameters for playing under other victory conditions, and probably no one else can either.
4. The system does not take into account secondary victory conditions.
However, I believe that the advantages outweigh the disadvantages, as I have repeatedly written above.

And in essence - I repeat. I don't care about the rating, so do as you see fit. After a dozen tournaments, something might change... If, of course, we wait... ;)

Well, I hope we'll wait for something. I'm currently pushing the Beastmaster in the tournament. But there are many viewers in the thread, but I think only 2-4 people are actually playing.

So, as I understand it, you suggest that my bonus should only be calculated for the linear component? But in this case, the result might turn out to be negative. Which, with an oscillator, would further reduce the final score. However, if you think about it, you are absolutely right. Since this bonus depends on the placement and the number of players, it should be applied to the linear component. But! It also takes into account the average rating of the tournament (according to your non-linear component), which directly depends on the strength of the players. And, therefore, it should be related to the overall rating, after the oscillator.
Although, it might be unnecessary, but, as I think, it reduces the drawback of the oscillator (with the halving of the last rating) by the fact