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

I already explained all of this in this thread. The position in the heroes is subjective if you start calculating several games with different compositions. But DNI is real – that’s what explains everything. So, what about a compromise? Don’t you want to help me out by fulfilling a small mathematical request?

Iron, if you need help, I will try to assist. If the goal is to create another round of the arms race, forget it.

It's difficult to switch from one system to another, even if the new one is more organized and disciplined – everyone speaks for themselves, for the sake of others. :)

P.S. I have to go to work today, in 5 hours. Until next time...

The position always corresponds to the rating: the higher the rating, the better the position...
But in Iron's system, the rating is the main factor, while in Amber's system, the position is. I won't come up with anything else (or they'll eat me alive:D)

I'm afraid you've forgotten the essence of the oscillator... In Iron's system, it's just the number of days. In my system, both the days and the position are equally important...

Exactly.
Here's another example (or rather, a counter-example): the same tournament, the same 50 players. The first round took 10 days, the next ones took eleven (or ten players took eleven days, seven took twelve, and the rest took fifteen, so you wouldn't say I'm giving an unrealistic example). What do we have? The players who finished in the same amount of time competed based on a secondary criterion, and someone, having put in the effort, showed their skill and outperformed everyone in their group. The question is, why shouldn't they be eligible for a bonus?

Compare the secondary criterion and give a bonus, but not based on RANK. I told you before that the part of your formula that takes RANK into account plays a dominant role. And in some cases, my contribution isn't even visible. I told you then that maybe you should add some kind of coefficient.

After all, all players will be sorted by rank, but 2nd place and 11th place will receive the same rating according to the Iron system. This is also unfair.

I've already said this. They played soccer, and it was a draw. It doesn't matter who ran more beautifully. The main criterion for victory is equal - a draw. Is that really so unfair?

The oscillator is the average between the rating by days (the Iron system) and the rating by rank... Two equally valid reference points have found a compromise in this method. Both rank and result influence the rating calculation... Explain to me, dissenting gentlemen, what is so controversial here? Please be objective. Phrases like "the oscillator halves the rating of the last place!" are not serious... This is not a minus, but an incentive to develop and not be at the bottom of the table... And in general, losing 2-3 points won't really bother an outsider, I think... 5 or 3 in the end, it doesn't make a difference. But at least we solve the problem of taking into account the secondary criterion. Iron doesn't take it into account at all. Which is more important?

We're going in circles. I gave you examples before of two matches with a skewed oscillator. Look for them. Or should I explain it objectively again? Do I have to mess around with a calculator again? Or search for these posts here?

But this is enough to call it a cooperative. Let's say they are average. But nevertheless, a part of the players who would be collectively ranked lower than them in the system. What's wrong with that?

This is a drawback of the system where placement is taken into account.

I'm afraid you've forgotten the essence of the oscillator... Iron's has only the number of days. Mine has both days and location...

It might turn out to be equal if you find the example that I was theoretically selecting. I can sometimes do it like Fireball too :) But you yourself said that the system needs to be tested for survivability with various examples, remember?

Added 2 minutes later

This is a drawback of the system where location is taken into account.

Yes. And it's present in your formula.

For me, the only and main criterion for calculating the rating is real results – days. And for Ambera, the subjective PLACE plays a decisive role in calculating the rating. That's what I don't like.

But the secondary criterion cannot be ignored. You stated that it is not important. But gathering more resources (units, experience, etc.) in the same period as your opponent – isn't that an indicator of skill? You should get a bonus for that.

Iron, if you need help, I'll try to assist. But if the goal is to create another round in the arms race, forget it.

No. I said I'd leave after this and won't interfere until you and Fireball provide your final formula. Is that okay?

Added 1 minute ago

But the secondary criterion cannot be ignored. You stated that it's not important. But gathering more resources (units, experience, etc.) in the same period as your opponent – isn't that an indicator of skill? You should get a bonus for that.

So, tie these indicators to a specific coefficient, BUT DON'T RELY ON THE RANKING. You'll immediately avoid the problem with cooperative play, definitely without any revolutionary changes on the forum. You took the easy way out. You just used the ranking, and that's it. It's easier than tying a bonus for a secondary parameter through some coefficients. If your formula's oscillator doesn't take the ranking into account, but does take secondary parameters into account, that would be good! In other words, you're considering secondary parameters in an easy way through a SUBJECTIVE ranking, but you're considering them in another way – not through the ranking

Exactly.

The second example (or, rather, a counterexample) is the same tournament, the same 50 players, the first one took 10 days, the next ones took eleven (or ten people - eleven days, seven - twelve, the rest - fifteen, so that they don't say that I gave an unrealistic example). What do we have? Among themselves, the players who finished in the same time competed according to a secondary criterion - and someone, having made an effort, showed a level and overtook everyone in their group. The question is - why can't he claim a bonus?

After all, all players will be sorted by place, but 2nd place and 11th place will receive the same rating according to the Iron system. This is also unfair.

I agree. The result is better, but the rating is the same. Unfair (although the best result among those with the same main parameter will have a more honorable second place).
Although this inaccuracy seems less significant to me personally (IMHO).

The oscillator is the average between the rating by days (the Iron system) and the rating by place... Two equally valid reference points have found a reconciliation in this method. Both place and result affect the calculation of the rating... Explain to me, gentlemen who disagree, what is so wrong here? Just objectively, please. Phrases like "the oscillator cuts the rating of the last place in half!" are not serious... This is not a minus, but a stimulus to develop and not be at the bottom of the table... And in general, the loss of 2-3 points for an outsider will not greatly affect, I think... 5 or 3 in the end, it doesn't matter. But instead, we solve the problem of taking into account the secondary criterion. Iron does not take it into account at all. What is more important?

The time limit is enough so that everyone is satisfied - and the organizers do not worry about possible fraud. And the players are not stressed out by sleepless nights. This is a minor problem compared to everything else...

I'm afraid you've forgotten the essence of the oscillator... With Iron, it's the number of days and only. With me, it's both days and place equally...

The oscillator is not the average.
More precisely: in the oscillator, the result affects the place, and the place, in turn, affects the rating. With Iron, it's the other way around - the result affects the rating, and the rating affects the place.
And in both cases, there are shortcomings, but in both cases, both days and place are taken into account.

Explain to me, gentlemen who disagree, what is so wrong here? Just objectively, please.

Objectively - the example in post 711.

I've said it before. They played soccer – it was a draw. Who cares if someone ran more beautifully. The main criterion for victory is equality – a draw. Is that really so unfair?

Well, it's not quite soccer here... There's a very clear objective: to score more goals. And in the tournament bracket, there are many secondary criteria for advancing from the group, and I think you know them. And the winner of a match can be eliminated from the bracket, while the loser remains. This maintains the intrigue. You want to remove it from the calculation. In soccer, running beautifully isn't necessary; it won't help. But knowledge of logistics will allow you to gather more resources (for example). There's a bonus for knowledge of the game. What does running have to do with it?

Added 4 minutes later

Yeah, and it's present in your formula.

Yes. I'm not denying that. So, should we abandon it because of this minus? Why aren't you looking at the whole picture? You're deliberately focusing on one thing – your approach is biased, and your opinion is questionable from an objective point of view... You need to evaluate all the pros and cons, so either talk about the system as a whole, comparing the pluses and minuses, or tactfully remain silent. I've already asked you to do this...

In both cases, there are shortcomings, but both consider days and placement.
Objectively - see example in post 711.

I agree - in my case, placement is automatically taken into account because the results by DAYS are always arranged in a certain order. And Amber then takes the SUBJECTIVE placement into account to consider secondary parameters, and that's where the mistakes are made, but at least the secondary parameters are taken into account. But who said that secondary parameters can only be taken into account through PLACEMENT?! I don't have enough mathematical skills to come up with a coefficient and take these parameters into account without using placement. But because of the mistakes, I don't want to.

Added 3 minutes ago

Yes. I don't deny that. So, should we abandon it because of this drawback? Why aren't you looking at the bigger picture? You are deliberately focusing on one thing - your approach is biased, and your opinion is questionable from an objective point of view... You need to evaluate all the pros and cons, so either talk about the system as a whole, comparing the pros and cons, or tactfully remain silent. I've already asked you about this...

How can I be more tactful? I'm telling you. You said that you use PLACEMENT to take into account secondary parameters. Take them into account without using the PLACEMENT parameter. Then I will appreciate your mathematical skills.

Added 6 minutes ago
A formula that will take into account only days (placement is automatically visible, showing who is in what position) and, in case of a tie in days, will take into account a bonus for secondary parameters (but without using the occupied placement for this), would be SUPER ACCEPTABLE, in my opinion. But how can this be done? Find a bonus that is not as large as it turns out when calculating with placement taken into account. Well, I showed it on the 4th. How a 28-point bonus is immediately deducted for second place compared to when placement is not taken into account. If someone else had also taken second place, a bonus could have been added to the 27 points in my system for the secondary parameter. But not 28. Because then it turns out that even if there was no one else in second place (there is no need to compare secondary parameters), a 28-point bonus is still added. It turns out that Amber is again hiding behind the secondary parameters, saying how good it is, but it's not quite like that. They choose examples that favor them. And here is an example if there are no identical results in terms of days. There, a huge bonus is simply added to the actual result for DAYS for the SUBJECTIVE PLACEMENT. At the same time, no bonus is given for the subjective 1st place for the actual result for days :) Amber, stop deceiving me :)

A formula that only considers the number of days (the location is already visible, so we know who is where) and, in case of a tie in days, takes into account the bonus for secondary parameters (but without using the occupied space for this) would be SUPER ACCEPTABLE, in my opinion. But how can this be done?

We've already discussed this before... We need to brainstorm; it's not something you can figure out right away. And there's no guarantee that it won't lead to new inconsistencies. But the more complex the method of calculation (and I think your suggestion would significantly complicate the scheme), the less transparent it will be for finding errors... we need to avoid making things even worse. After all, I've already tried to link this before

We've discussed this before... It requires some serious thought, as it's not an easy issue to resolve. And there's no guarantee that new inconsistencies won't arise. But the more complex the calculation method (and I think your suggestion would significantly complicate the scheme), the less transparent it becomes for identifying errors... it could end up causing even more problems. I've already tried to link these things together in the past.
Alright, I'll think about it again... Maybe something new will come to mind.

Okay, fine. But this would be better. And why do I call the RANKING subjective? If there were other strong players in the 4-player tournament, Fire's ranking would shift downwards. But if the RANKING wasn't taken into account in the formula, Fire's result relative to the leader wouldn't be affected; it would remain a real 27 points. But with the ranking taken into account, its rating would drop. That's a bummer :( and it turns out that the ranking is subjective, depending on the players present, and Fire's rating could jump from 55 to who knows where down. But without taking the ranking into account, it's clear. No matter where it's moved in the ranking, it remains firmly at its 27 points, which depend on its actual result in the days. Fireball should also think about developing the formula in this direction together with you.

And considering the placement, his rating would have dropped. That's how it is :( and it turns out that placement is subjective based on the current pool of players

We are going in circles. Well then, I will repeat myself.

In a situation where a strong player enters a tournament and raises the bar by taking 100 points for a super-run, all other players will be left picking up miserable crumbs of rating. This is an alternative scenario with a non-zero probability. The effort spent on the tournament would be wasted... And the place taken by the second player could have compensated him for the loss of rating when playing against a dinosaur... After all, he still beat everyone else and deserves a reward (grails not counted). But in the end, he will get only 5-10 points more than an outsider who sent their save simply by mistake... (a conditional figure, but a completely possible situation).

That is also a question. And we are not talking about the probability of the outcome here. We are talking only about the flaw in the calculation mechanism.

We are going in circles. Well, I'll repeat myself.

I'm going in circles because of you. But today is the last lap and then a break — I need to see if anyone even wants this besides us.

In a situation where a strong player enters a tournament and raises the bar by taking 100 points for a super-walkthrough, all other players will be picking up miserable crumbs of the rating. This is an alternative scenario with a
non-zero probability. It would be a pity to waste effort on the tournament... And

Maybe we should stop guessing. Look at the ratings for ACTUAL past Kanigem stages (by my system and by your system) published by Alex Pank. Almost all leader walkthroughs there are super-walkthroughs. Do you disagree with this? However, everything is fine. If you compare the outsider ratings across stages and the general rating in those three tables, they only have miserable crumbs under your system. More accurately, their ratings are about half as much as under my system and Alex Pank's system :) Amber, why say silly things — people aren't blind; they can look for themselves. Even if it were as you say regarding my system (BUT YOUR SYSTEM SUFFERS FROM THIS GLITCH AGAIN, NOT MINE :) — look at the real rating tables :), then I see nothing wrong with players receiving ratings corresponding to their results. And if a leader lowered the bar, meaning second place showed a poor result in days, I don't believe a player should receive free food stamps in the form of a 28-point bonus when they actually earned 27 for their real result (the bonus is more than what was actually earned in the tournament). Moreover — placement is subjective — and your free stamps in one of the tournaments could subsequently distort the entire rating and results of the WHOLE championship. Why should an outsider-level player receive bonuses just because mid-tier players failed to show up? Some outsiders will be lucky enough to participate in a "freebie" tournament, while others won't. Again, we are heading toward rating imbalance.

the place taken by the second player could compensate them for the loss of rating when playing against a dinosaur... After all, they still beat everyone else and deserve a reward (grails not counted). But in the end, they will receive only 5-10 points more than an outsider who sent their save simply by mistake... (a hypothetical figure, but a very possible situation).

You're exhausting me with your compensations. All the flaws in your system are due to accounting for placement; I'm tired of repeating it :) Now I'll blast out a treatise with an example of who beat whom and so on :) You're just pulling justifications for this parameter out of thin air.

That is also a question. And we are not talking about the probability of an outcome here. We are talking only about the flaw in the calculation mechanism.

That's exactly the point, especially when they start using placement to calculate ratings in OFF mode. The calculation ends up not just flawed — that would be one thing — but BROKEN. And that's no joke :)

Added after 3 minutes
I still want to explain once more why I dislike rating systems that use a parameter such as the player's placement in a tournament in their calculations. Such rating systems might be suitable for some sports, but certainly not for Heroic OFF tournaments. Take chess, for example. Before competing in a tournament, there is a player selection process — a bunch of competitions. In selections and tournaments, chess players fight each other PERSONALLY, and to break higher in points, they must personally achieve victory over every specific opponent currently in their path. Therefore, in the end, when they've sorted out their PLACEMENTS, calculating their rating could quite well happen via a system that considers placement. It's unlikely you'd encounter such a flaw as a lack of strong players in such tournaments :) A vacant spot is never left empty for long. But look at Hero OFF tournaments. The roster varies from tournament to tournament and across different game versions — there is no selection of players for the next tournament. Moreover, in an OFF tournament, you don't have to fight everyone — rather, everyone fights the map individually, and no one interferes except stupid computer heroes. Placement is determined simply by the main victory criterion on the map from a starting save (for example, speed — DAYS). A flaw such as the absence of strong players in a tournament can easily occur. Furthermore, the number of people in a tournament is not limited. In chess, it is. Therefore, there are practically no weak players there, or their number is limited. Now, I'm not talking about the total absence of strong players in an OFF tournament (that's a special case), but say one person came to the tournament and five such players didn't. Or mid-tier players could simply be absent. Under such random circumstances, situations arise where known weak players can suddenly take higher places purely due to the absence of strong or mid-tier players, and under a rating system with PLACEMENT, they will receive a rating inflated relative to their real result. Another situation — perhaps the opposite happens: an influx of strong and mid-tier players. Then weaker players occupy a place far different from what they could have taken in their absence, and their rating under such a system will be understated relative to their real result in days, even if they play consistently and their real result in days relative to the leader hasn't changed. Therefore, I call a parameter like placement SUBJECTIVE, but the result obtained by each player from the map from their starting save in days is REAL. Why should such a subjective parameter as placement have a GREATER influence on the rating under certain circumstances that occur randomly? In such a state of affairs, a rating system considering placement can easily mess up, and you probably couldn't account for all situations. Let's look at a real example from a stage in 4. Here it is very clearly possible to show how this flaw occurs. And may Fireball forgive me. But I'll say right away that I have nothing against him personally. On the contrary, I'll emphasize that he did great by completing the map. And not just him, but all players at this stage. And I hope they don't regret completing that map and I think they increased their experience in this version of Heroes. But nevertheless, we have to examine the flaw using this example — I haven't found another real example yet.
Axe - 3 days - 1st place - 100 by Axe - 100 by Amber
Fireball - 11 days - 2nd place - 27 by Axe - 55 by Amber
.
.

We see that under a system that does not consider placement in the calculation, Fireball receives a rating of 27 for his REAL result in days. Under a system considering placement, he receives 55. The bonus for SUBJECTIVE placement turned out to be even GREATER by one unit than the points for the real result in days. What a bonus — 28 points. And for what? Amber explained to me before that this bonus encourages the fact that a player beat a bunch of people, and there were strong players among them; surely he has the right to receive a bonus. He does. But not like this and not so much. And I tried to explain this flaw to him. In this example, let Amber see for himself which "strong" players Fireball beat and how many of them he beat. To tell the truth, there simply weren't any strong players or even mid-tier players in 4 — and there were few people overall. Fireball is a newcomer to 4 — he honestly worked through the map and got his honest REAL result in days. But circumstances happened such that there were no more strong or mid-tier players. And he subjectively took SECOND place. And for these RANDOM circumstances, AmberSoler adds a 28-point bonus, and Fireball receives a total rating of 55, corresponding to a mid-tier player, although he is not yet one in 4 (he trailed the leader, even if also subjective, by three and a half times). Who disagrees with this statement? Speak up, you monster :)
Suppose, what would happen if more strong and mid-tier players had appeared at the tournament. They would have pushed Fire down n-positions so that under Amber's system he could easily receive a rating LESS than 27 (which is his honestly earned rating for the result in days). On what basis? Why is he punished under such circumstances? He's a newcomer — he hasn't been able to beat the mid-tiers yet; does he have to jump out of his skin — he'll get his chance :) So we end up with distorted ratings not only for Fireball, but also for the other players following him. If anyone didn't understand, I'll note that for players at the bottom of the table, Amber's placement bonus works in reverse: not toward increasing from their real result in days, but toward decreasing from their real result in days.

In this example, it is clearly seen what flaw exists in Amber's system and it's clear that it doesn't exist in Axe's system because I don't include subjective PLACEMENT in the formula. Perhaps they will still struggle to eliminate this flaw. This could be done by introducing a complex parameter, such as TOURNAMENT STRENGTH. However, what would this parameter do? For the most part, it would have to eliminate the flaw introduced by such a parameter as accounting for the player's SUBJECTIVE placement. I just made it simpler. I threw out the PLACEMENT parameter; accordingly, I don't need the TOURNAMENT STRENGTH parameter either (which is quite a hassle).
---
However, it's clear that both systems have a flaw regarding the SUBJECTIVITY of the leader's own placement. Let me explain. If Axe hadn't been at this stage, Fireball would have received a rating of 100, whether by Axe's system or Amber's system, and the ratings of other participants in this tournament would dance toward inflation. Nothing can be done about this. After all, some starting point was needed for the calculation, and that turned out to be the Leader's result, relative to which the ratings of other players in the tournament are calculated. I have suggested two solutions so far (they aren't perfect), but no one wants to work in this direction, although as I already said, all existing systems currently suffer from this flaw.
way 1 - introduce acceptable limits for completing the map for the leader himself and apply a corresponding coefficient. Define acceptable limits in a sufficiently wide range. Use a parameter such as map size for this purpose. In short, the essence is this — if the leader's result fits within certain frames, for example on an S map - 2 weeks, M - 3, L - 4, XL - 5, then the coefficient is 1. That is, the leader takes his hundred. If the leader doesn't fit the limit, he will take a rating less than 100. Others dance from the leader's result, as we remember. However, it's clear that here, with the acceptable limits I specified, Fireball would still have received 100. That is, some jewelry-like tuning of acceptable limits is needed here. I don't know how to do it, and the numbers are just pulled from the ceiling based only on tournament experience. And who knows how to accurately define these limits?
way 2 - introduction of a virtual player. But it seems we proved that there would be even worse flaws. Although I still think mathematicians should rack their brains — maybe something would come of it. Maybe introduce the virtual player in some other way than what I proposed.
In short, in my system, only the 1st way with crude tuning currently sticks out. Other systems don't even have that. There, even if a weak player takes a SMALL (easy) map in a thousand days and becomes the leader — they grab 100.
---
The drawback of Axe's system is considered to be that it doesn't account for secondary parameters for determining the winner (like who gathered more mercury when days are equal). At the same time, this is considered an advantage of Amber's system, as if the system accounts for it, but how? How does this advantage happen. It happens through the parameter of SUBJECTIVE PLACEMENT. Well, the flaw with subjective placement was discussed above. And it turns out that it's hardly worth trying to separate the ratings of players who have identical completion days but different amounts of mercury at the cost of such a flaw. Look at the example from the 4th stage. No one there has identical results (except for the last two places), but the bonus pushes for second place. In other words, Amber's bonus pushes PURELY for placement, not for secondary signs. As proof of this, one can further cite the fact that if at the bottom of the table there are several players who showed an identical result in days but have different amounts of mercury, none of them RECEIVES any bonus for secondary parameters because their rating still decreases relative to their real result in DAYS. They receive even less than they would under my system even without this bonus :) With Amber, the player who has more mercury simply has a rating that DECREASES LESS relative to his real result in DAYS :) No, well actually I agree that since it decreases less, in principle, this can be called a bonus, if you have the heart to call it that :)

Conclusion: Agreed. It is desirable to include secondary victory conditions in the rating calculation formula, but this should NOT be done THROUGH the SUBJECTIVE placement parameter. And the bonus there should not be very large — it's intended only to separate the ratings of players who finished in the same number of days, not to pull one of them sharply up in rating or push them down if it happens at the bottom of the table, since the PRIMARY parameter for these players is still THE SAME (we've already seen what kind of bonus can result under certain circumstances). Who disagrees with this statement? Speak up, you monster :)
---
In recent discussions, another term cropped up: "co-op". If such a circumstance occurs in a tournament and a co-op player wedges themselves into the final table between the leader and the mid-tiers, then the rating considering placement will again mess up specifically. Axe's system handles this situation with dignity. However, both systems will glitch if a co-op takes the leader's place. But I assert that the flaw in this case in Axe's system will be much smaller than under Amber's system. Who is against this statement? Speak up, you monster :)
You can experiment with calculations yourself, just consider various situations and convince yourselves of this. Read the posts above in this thread on this matter and you will all understand without difficulty regarding this flaw. Revolutionary ways to eliminate such a flaw were proposed by Amber. It'll be quite a hassle. And I'm still not sure that even through such a way the situation with co-ops will be completely eliminated, but it may succeed in reducing the probability of their appearance if Amber can carry out this revolution.
---
Request to Amber: no offense, I tried to more or less OBJECTIVELY summarize the results of our personal dispute by using a real example from the 4th stage of Kanigem. Remember, the imbalance introduced in the 4th stage tournament when calculating ratings under your system could affect the results of the whole championship when summarizing totals for the Universal nomination. Why? Because it turns out no one knows exactly how the totals for this nomination will be summarized.

And what else I want to say is that your oscillator (your part of the formula) is good if my part of the formula is thrown out of it :) AND APPLIED to calculate ratings for ON-LINE tournaments :) Seriously. You wanted objectivity and impartiality. Well, here you go; take it and sign for it. My part of the formula doesn't rule there at all. Because in ON-LINE, the days parameter isn't important. What matters is the personal victory of one player over another. In other words, PLACEMENT really rules here because the higher it is, the greater number of opponents a player actually defeated in a personal duel, meaning they actually beat them in this tournament. Days can only play the role of a secondary parameter to identify the best if some ON participants end up with equal points according to your formula; however, for this, my part of the formula doesn't seem to fit. So, organizers of ON-line tournaments — don't pass by if you want to introduce a rating system for the participants of your tournaments.

That seems to be all for now. Now I think it will be clear to everyone that I argued against your system not because I didn't understand the essence of the oscillator and your formula. Rather, the opposite: I understood it very well and worked through several flaws with examples that I don't have. The same flaws that I have are also present in yours, sometimes even in an enhanced version. The drawback of my system is that secondary parameters (mercury) are not taken into account. But unfortunately, accounting for these parameters in your formula is not a virtue because they are accounted for through the PLACEMENT parameter.

I think if anyone can manage to read this article carefully from start to finish, they will understand much about the proposed rating systems and see their advantages and disadvantages.
---

And a request to those who read to the end and understand something — speak up, whether you agree with my statements in this post or not. Don't be shy; as for the monsters, that's a joke :) Otherwise, I don't know if I should continue defending my positions on these issues or not. If many consider them wrong, then why would I continue to argue. But if no one responds at all — it means nobody wants this, so why should WE argue :)