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

I will try to summarize all the suggestions made.
1. A purely game-based rating is calculated using the system proposed by the respected IronAxe, without taking aging into account, but with a minimum quota of tournaments per year (I suggest 6 tournaments - once every two months). The rating of those who have played in fewer tournaments can be aged.

I think I've taken all the suggestions into account. Any comments?

I was always talking about the CURRENT one. But the question is - if there is a championship, can't I duplicate the CURRENT Player Rating of those who participate in it?

... but we just need to somehow start taking into account the current ratings of the players on this resource.

You are not taking into account the number of participants in the tournament and the player's position in the tournament standings. That is, you are simply accepting Option #1 from my post, which was based on IronAxe's suggestion.
I ask you to consider that the oscillator primarily reflects the player's strength in relation to other participants in the tournament.

Perhaps we should differentiate between the concepts of RANKING – which depends solely on the speed of completion, and GRAILS – which depend on the number of participants, and can also be spent on the doll?

Perhaps we should differentiate between RANKING – which depends only on the speed of completion – and GRAILS – which depend on the number of participants, and which can also be spent on the doll?

When the second component artificially elevates an outsider in the ranking, it's not quite a fair comparison. The placement in the tournament is secondary. What matters most is the time it took for a participant to complete the game and how close they are to the leader. Therefore, comparing the strength of players in this way is not entirely accurate.
I used a STUPID example to illustrate this and compared two stages. In one case, a player who is a thousand times behind the leader takes second place and, consequently, receives the same rating as a player from another tournament who is eight times behind the leader but also takes second place! IS THAT RIGHT?

Simply, if your proposal is suitable for calculating grails, then why not use it? Ideally, these two concepts could be combined, BUT grails are spent, while the rating should remain.
Maybe we will arrive at such a solution?

Your "silly" example is quite appropriate, however, a third participant must be introduced to complete the picture.

Hence the question: if a third participant completes the map in 500 days, then the outsider with their Walkthrough in 1000 days will be third out of three, not second out of two.

And if the newly introduced third participant completes the map in 1200 days, then the one who completed it in 1000 days becomes second out of three, not third... Moreover, their result remains unchanged, but the rating will depend on the play of ALL participants, not just the leader... This is not the same thing.

And the oscillator is designed to take into account all tournament participants (not only their number, but also their relative positions). The second parameter, which you have complaints about, is responsible for this point. The second (linear) parameter is not applicable on its own, but in combination with the first, it gains meaning... Don't rush, just experiment with the numbers...

Your "silly" example is quite appropriate, however, a third participant must be introduced to complete the picture.

Hence the question: if a third participant completes the map in 500 days, then the outsider with their Walkthrough in 1000 days will already be third out of three, not second out of two.

And if the newly introduced third participant completes the map in 1200 days, then the one who completed it in 1000 days becomes second out of three, not third... Moreover, their result remains unchanged, but the rating will depend on the play of ALL participants, not just the leader... This is not the same thing.

And the oscillator is designed to take into account all tournament participants (not only their number, but also their relative positions). The second parameter, which you have complaints about, is responsible for this point. The second (linear) parameter is not applicable on its own, but in combination with the first, it gains meaning... Don't rush, just experiment with the numbers...

The concept of an oscillator, as a parametric index reacting equally to changes in several unrelated parameters, was not invented by me. This mechanism has been tested and adopted by all methodologies researching the behavior of randomly changing digital sequences (tournament results are exactly that). So, in the first approximation, everything here is correct, do not doubt it. And to justify this to you, I would have to go beyond the scope of the topic :) Maybe it's not worth complicating things?
And for calculating grails, this mechanism is certainly applicable.

Yes, I understand that, but why do you deny that the second parameter can almost completely absorb the first – it’s clear from the example, and we are trying to get a GENERAL CURRENT game rating for all participants of the resource playing in tournaments :)

In my case, the situation becomes terrible if the leader of a certain stage completes a map, for example, in 1000 days, and the rest do even worse, then, due to the connection to the leader, the results will still be incorrect. In this case, I can only say in my defense (and the leader will say the same). And you would try it yourself, maybe you would complete this map even worse :)

I agree that it might be worth adjusting the importance of the second indicator, i.e., applying a constant (or perhaps a variable) decreasing coefficient to it. I just don't have any ideas about what that coefficient should be... I won't pull it out of thin air, and I don't know how to justify it yet...

You know, I've always thought that the rating is calculated as the sum of the results of different unrelated tournaments. It is unlikely that there are methods to directly link the results of one tournament with the results of another. The main thing is that all participants in each tournament are in different conditions... Take the triathlon - how can you link a cycling race and swimming? In no way. But as a sum of results (the speed of completing a stage), it can be linked. The main thing is that all participants complete all stages under equal conditions...

That's what I'm saying – it's all about the coefficient again. Perhaps it's the strength of the tournament, based on the CURRENT player ratings, which no one has yet.

The comparison isn't quite appropriate; they're not swimming or racing, but clicking the mouse and keyboard, using their brains, closer to chess.

You know, it always seemed to me that the rating is calculated as the sum of results from different unrelated tournaments.

Damn, why did you delete some of the formula's markings? Check it against my simple example, comparing it with the two stages and different contingents. The examples are above in the thread; there's your normal example for one stage and my silly one with 1000 days for the other.

That’s exactly what I’m saying – it’s all about the coefficient again. Perhaps this is an issue with the tournament’s strength being based on the CURRENT player rankings, which no one has yet.