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

to IronAxe: You clearly misunderstood me...
I meant that this figure will be the minimum in the second column... i.e., in the second column, the figures will be from 100(leader) > X(n-th player) > the outsider's figure from the first column...
Anyway, to make it clear, I'll make a table and post it in this thread... and we'll see from there... :D
Regards.

I don't know how I understood you, but the situations described in the previous post are real under the layout you are proposing. When using the oscillator, you will find that the player in second-to-last place in the rating ends up lower than the player in last place. After all, are you suggesting that only the player in last place transfers their rating from the first column to the second ?? If you make tables - also check the option where the rating in the first column for the second-to-last person differs from the rating in the first column for the last person by approximately 0.5.

to IronAxe: well, what else can I say... read more carefully... :D

i.e., in the second column, the numbers will be from 100 (leader) > X (n-th player) > the number of the outsider from the first column...

I think I expressed myself clearly...

But to avoid talking nonsense...
Look, here are the data for the second stage... with a small upgrade to the oscillator...
Take a look and enjoy... :D

Sincerely.

to IronAxe: well, what else can I say... read more carefully... :D I think I expressed myself clearly...

No, this line is unclear to me, and the formula is too complicated – I don't really want to delve into it. BUT, based on the results in the table in the "HOW IT BECAME" column, I like it more than a simple oscillator, and with the bonus and a 5-unit threshold. By the way, when I was arguing with AmberSoler, I told him that if he came up with something that would make the transition from the first column to the second smoother, so that the second component wouldn't suppress the first as much in our formulas as it did in his version. But we didn't go any further. If you really like complicated formulas and taking up space into account, I would agree that the main official rating should be calculated using the system that Kyle has now proposed. I will leave my system as an alternative – it shows the rating purely based on speed and does not take into account any secondary factors (buffs, mercury), and therefore does not take into account the space occupied. If AlexPank wants to keep both systems, I'm not against it. But sooner or later, the underdogs will catch up with him and beat him :) For the bonus. Just kidding :) Let's have a chess federation! ;)

It results in a rather complex formula. Plus, you can't divide by zero (and the first number turns out to be exactly that). Overall, it's not bad, though. But the results are too close together; the leaders, in my opinion, deserve a greater advantage.
If you want, you can calculate using this formula, but I won't be able to handle it anymore.

The formula is quite complex.

Calculating with a bonus that takes into account the results of previous games is more difficult. Moreover, I am sure that there will be, or already are, errors there (possibly not due to you), but it is unlikely that anyone will undertake to recheck. In the results tables from Tournaments, player nicknames are sometimes distorted, and it is difficult to understand whether it is the same player or different ones. Consequently, if the nicknames are different, you will calculate for each one, and then, taking into account the rating of previous games, you will calculate further Tournaments. And after 5-10 Tournaments, some guy will show up and say - listen, the player ***DeMoN*** and DeMoN are me, and you are counting me as different people. And here are two options: 1. You send him away and don't recalculate anything (however, the rating is already distorted for EVERYONE who played with this player in the Tournaments, and if there are not one, but two or more of these guys?). 2. You undertake to recalculate all 10 Tournaments, but then it will be easier to hang yourself :)
Therefore, all other systems are advantageously different from the system with a bonus because they do not take into account the results of previous Tournaments to calculate the results of the current Tournament. In the situation described above, I don't need to do anything except add the results of ***DeMoN*** and DeMoN into one slot in my summary table. Problems may arise with aging. However, in this case, as in your first option, I can send this guy away, since he did not keep track of his rating, or I will only have to deal with him, since his past results are not tied to the results of other players who played with him in the Tournaments, and I do not need to recalculate their results. Do you understand what I explained?

Plus, you can't divide by 0 (and the first number turns out to be exactly that).

I don't understand, how does the first number turn out to be exactly that? And what is it equal to when you divide by 0? :) Kyle, explain this moment of your formula to us.

But in general, it's not bad. But the results are too close, the leaders deserve, IMHO, greater advantages.
If you want, you can calculate using it, but I won't be able to do it anymore ;)

The leaders deserve what they get from their result. Namely, always 100. And if, in addition, their result is very outstanding (that is, they are far ahead of everyone else in terms of days and even far ahead of the player in 2nd place in terms of days), then they automatically reduce the rating of the others, depending on their results. This is the bonus. Everyone gets according to their result. And if the results of the players are close, then I personally don't see anything wrong with that.
Therefore, I suggest that you calculate Kanneghem's Tournament as you are calculating it now, so as not to recalculate anything (according to all three systems).
And in the future, adopt Kyle's method and calculate according to it.
Get rid of the system with a bonus, and in the future, you may have less of a headache ;)
I still think that ***DeMoN*** and DeMoN are the same player, so an error has already crept into your rating with a bonus, which may turn into a snowball in the future, when it will be impossible to correctly recalculate the distorted results of all players, or you will have to spend a lot of personal time on it.

Fixed the issue with "Demons." I'm letting you know that this didn't affect the Clan Wars tournament in any way. Therefore, I'm not going to abandon the existing systems. Regarding 0: Did you see Kyle's formula? The leader's result is derived from a formula where the first-place score is subtracted from itself in the denominator, resulting in 0.

I fixed the issue with the "Demons." I'm letting you know that this didn't affect the War of the Kingdoms tournament. Therefore, I'm not going to abandon the existing systems.

Well, I just pointed it out to you in time. You don't want to admit that such errors might occur in the future and might not be noticed in time. Please fix the Kolbazzer issue as well – there are two of them with different spellings. And by the way, Kolbazzer himself mentioned in one of the threads that his nickname was distorted in the tables, and his rating was also calculated as if he were two different players.

Regarding 0. Did you see Kyle's formula? The leader's result is obtained from a formula where the first place is subtracted from the first place in the denominator, i.e., 0.

:) Figure it out yourselves; I've expressed my opinion, so let him come up with a way to fix this issue or explain his version. The results according to his system look better and are more fair to both leaders and underdogs compared to systems with a simple oscillator and a bonus with a threshold. My system is fundamentally different from all three, as it doesn't take into account the position, so it makes sense to keep it as an alternative. If you're not too lazy to calculate using those systems and you're not afraid of headaches, then I wish you good luck :)

Regarding point 0. Have you seen Kyle's formula? The leader's result is derived from a formula where the first-place value is subtracted from itself in the denominator, resulting in 0.

Forgive me, but: You made me laugh...
And how is the second column of the oscillator calculated for the leader?
It's the same zero... :D

To make everything look good and avoid "division by zero," you simply need to assign 100 points to the leader by default... and use the formula for the second and subsequent players...

Sincerely.

Yes, I understand that perfectly. I'm not against your system. I'm just tired of having to keep track of four different versions. If you decide to take on this task yourself, I'll be happy to see the results.

Yes, I understand that perfectly. I'm not against your system. I'm just tired of having to keep track of four versions. If you decide to take on this task, I'll be happy to see the results.

We could certainly try... :rolleyes:
But I have a feeling that we'll soon be drowning in a sea of tables and numbers... :smile33:

Yes, I understand that perfectly. I'm not against your system. I'm just tired of having to calculate based on four different versions. If you decide to take on this task, I'll be happy to see the results.

I suggested that you calculate the results for the Kanigem tournament in the same way you usually do. To summarize the results for Kanigem based on your calculations. But after the tournament, we can switch to the latest version proposed by Kyle. It's simpler compared to the calculation with a bonus that takes into account the previous results of the players, and it's better than a system that simply uses an oscillator for Amber, where the second component of the formula negates the first. Besides, there's no point in looking at the results; everything is already clear.

I recalculated using the new system... take a look... the top 30... Below is the complete table... By the way, I used the table of the respected AlexeyPank as a basis; I hope he doesn't mind...

General rating for all games
Place	Nickname	Total	Number of games	Average
1	Zingel :smile49:	409	5	82
2	IronAxe	406	6	68
3	VDV_forever	377	6	63
4	vbn	308	4	77
5	AlexeyPank	290	6	48
6	Sage	279	5	56
7	Qwerty	276	3	92
8	Kyle	251	4	62
9	Alex_R	211	3	70
10	Dmitriy A S	200	2	100
11	Sergh	187	4	47
12	Lexx	171	3	57
13	Richard I	164	2	82
14	OlegRS	144	5	29
15	FireBall	141	2	70
16	Inquisitor	139	2	69
17	YoshiDo	119	2	60
18	vbs	102	2	51
19-20	Shoot-nick	100	1	100
19-20	Prorock	100	1	100
21	Geksorin	99	2	50
22	fanat	99	2	49
23	Radik	94	1	94
24	DeMoN	92	2	46
25	Bandit	91	1	91
26	Grib_OK	90	2	45
27	K@lb@zzer	90	2	45
28	Ataman	87	1	87
29	Maximix	87	1	87
30	Mikki	82	1	82

Sincerely.

Yes, by the way, Grib_OK and Tokha are (if my memory serves me correctly) the same person.

The results are interesting, but debatable. I personally didn't expect to see myself in such a position. Actually, I should be ranked lower, much lower. Apparently, even this ranking isn't a perfect solution.

The results are interesting, but debatable. I personally didn't expect to see myself in such a position. Actually, I should be lower, much lower.

But why?
Since this is a cumulative rating, the more games you play, the higher your rating... it's all correct... you're the last one among those who played 6 games... I don't see anything wrong with that... because the overall rating combined with the average rating shows who's who... :D

Sincerely.