Posts from Ideas and suggestions.

I believe the coefficient was introduced to prevent rating inflation on quick arenas and M200.
It could be limited by the formula: min(K, K_xl+u=60).

What exactly is being proposed? To make XL+U templates be valued the same as XL+U?

Yes, at least like that. And ideally, it should be based not on size, but on the template. K-values on mirrored templates (especially on x3dm) should, in my opinion, be the highest, because these are the most skill-based templates.

I disagree with Lizard regarding the revision of the rating system
1) A game on the G+U template takes longer regardless
2) It promotes additional popularity for a previously underused map size
3) I am confident that you can play 2 games on Mirror or Jebus in the same time as one G+U and get the same number of points.
Therefore, I believe the current formula is correct

p.s. Kostya, we are not talking about balance or skill now; we are talking about a different format, a different strategy, and no one is forcing anyone to play a coin flip.

I disagree with Lizard about revising the rating system
1) A game on the G+U template takes longer regardless
2) There's additional popularization of a previously unrequested map size
3) I'm sure you can play 2 games on Mirror or Jebus in the same time as one G+U and get the same number of points.
Therefore, I believe the formula is currently correct

p.s. Kostya, we're not talking about balance and skill right now, we're talking about a different format, a different strategy, and nobody is forcing anyone to play a coin flip.

1) No, it doesn't take longer. Except maybe games with your participation.
2) Popularization of WHAT??? And most importantly - why?
3) So you equate g+u, mirror, and Jebus? Speechless.

p.s. Vasya, I'm talking about Heroes of Might and Magic. And you, forgive me, about nonsense.

I don’t want to get into a debate about which template is better or worse, more skill-based or more coin-based, but regarding playing 2 games on Mirror and getting the same number of points as on G+U, Vasya is indeed being disingenuous — the difference in coefficients there is at least 3 times, if not 4...
And what does "popularize" even mean? What does an inflated rating coefficient, which logically should reflect the level of play, have to do with it? If a template is interesting to people for whatever reason, they’ll play it even without extra coefficients. But as it is, the majority of players are only playing it, by your own words, with the idea of getting more PTS if they win. Let’s go ahead and popularize some other templates then — why stop at such a small coefficient? Let’s introduce some new template that gives 1000 PTS for a win; win once and you’re top 1 — popularization! :smile03:
PS Although I was wrong about the size of the coefficients, as I was told, it turns out like this: For Mirror K=45. Jebus Cross 40, H+L+U 60, G+U 105. But that doesn’t invalidate everything else I wrote.)

I disagree with Lizard regarding the revision of the rating system:
1) Playing on the G+U map always takes longer.
2) It promotes the popularity of a map size that was previously not in demand.
3) I am confident that you can play 2 games on Mirror or Jebus in the same time as one G+U game and get the same number of points.
Therefore, I believe that the formula is currently correct.

P.S. Kostya, we are not talking about balance and skill right now; we are talking about a different format, a different strategy, and no one is forcing anyone to play on G+U.

1) I took a quick look at your Twitch VODs, and out of 20 streams, the longest G+U game took you about 5 hours to finish, and the fastest one was about 2 hours. I'd say the average would be 4 hours. I think we all know how long it takes to finish an average XL+U or H3DM game, so what you're claiming is complete nonsense.
2) This argument is also nonsense. Why should the developers forcefully try to popularize what is frankly completely unsuited for competitive games by awarding additional rating points? People are going to play what they want to play. If the bonus rating points are the main reason why you keep playing on this map size, then maybe G+U isn't that great. It just artificially inflates the rating for no actual reason.
3) https://puu.sh/EaYBb/7ac1bee39c.png - a random example from heroes.3tv
6 hours, 2 games, +150 points. Good luck getting anywhere close to that number on Jebus, Mirror, or whatever you want. How are these fluctuations in rating even remotely healthy for the scene? You can jump 300 points in as many as 5 games; it's completely ridiculous.

Im sorry
but

In this case, the size doesn't matter. If a standard game on XL+U and G+U takes roughly the same amount of time, then why should it award a different amount of rating points, and why is the difference so significant? :confused:

And if tomorrow a G+U template appears where the game ends in 30 minutes? Would that be okay? :eek:

In my opinion, after XL+U, the coefficient shouldn't increase, or should increase very slightly. (for example, if XL+U gives +20 points for a win, then G+U should give around +30).

Yeah. It's just ridiculous to watch. When some people grind out points by the handful/dozen (playing against +/- equal opponents), while others literally abuse the rating system, gaining one-tenth! per win. (If you imagine a similar situation on HeroesWorld based on the current rating: FTS beats Lucky on G+U and gets 1/10 of the difference between 2k and 1.4k-1.2k, that is 60-80 rating (in reality +20). A rough estimate. Yes, the rating systems differ, but I've accounted for a decent margin of error.)
And don't tell me the game went on for 10-12 hours yesterday, that's just an isolated case. Such cases happen on HL+ and mirrors too.
It's so obvious that well...

And when will starting rating finally be added? Without it, the top grows at a cosmic speed.
Combined with these G+U games, accounts can be boosted just way too easily.

By the way, a certain Messiah, who is currently in 7th place, has 56 (!!!) draws. He's simply the "friendliest" player of the year.

I entered the formula Sav gave me for calculating lobby rating and made a table.
On the left is the rating difference between opponents. The sum of what would be given for a win and taken for a loss is always equal to K according to the formula. With equal rating, they give and take K/2. (105 for G+U, 60 for XL+U, 40 for Jebus Cross.)
On the right is how much rating they will give for a win and a loss at a given difference on different map sizes + in the last column the coefficient K=34, which seems to be the case for HW (I checked this on several examples; if anything, correct me).

What is visible from the table in the tab:

1) With a rating difference of 110, a player with a higher rating receives 2 times less than the opponent in case of a win (accordingly, loses 2 times more).

2) When people with a difference of 146 play, in the case of a win by the lower-rated player on G+U, they become equal in rating!!!
For XL+U: difference of 72
L+U: 52
Jebus Cross: 46
M200: 21

Conclusion: the K coefficient is unrealistically inflated (not just G+U compared to others, but overall for all templates)

Either everyone needs to gain more significant rating numbers, then these jumps would look adequate (in that case, newcomers would need to spend an unreal number of games to even get close to the top/longer-playing players). But still, there should not be such dominance of G+U over the others

Let people play G+U; it looks very spectacular. In general, the idea of making the number of awarded PTS dependent on the duration of the match isn't that bad. It should still be easier for newcomers to gain rating, and harder for top players to maintain it. So, the fact that matches last longer for newcomers isn't that bad.

I apologize, but where did you see a dependence on the duration of the game? It only depends on the size of the map and the difference in player rating.

Let people play G+U, it looks very spectacular...

That's right. And let them go to Korzh's concerts. And let them eat at McDonald's every day. And let them die on the day they retire, wearing a T-shirt with the print "... - our President!".

Conclusion: the K coefficient is unrealistically high (not only G+U compared to others, but also in general for all templates)

Quite a strange conclusion. The coefficient is simply multiplied by all rating changes, so if the K coefficient is reduced by half for all templates, then the rating growth rate for all players will also decrease by the same factor of 2. In essence, one rating point will be worth two now. It will be equally difficult for newcomers to catch up with the top players, whether in the case of the top players gaining twice as many rating points with the current system, or in the case of the current rating values of the top players with a system with a K coefficient reduced by half.