Player rating for tournaments across all installments of Heroes.

sigh.. naive... I know one person who I think would want to argue... :D

I won't argue much :) I have very little time right now. The sixth stage of Kanigem is ending - and I haven't even downloaded the map yet :( I've abandoned my alternative Rating, I can't get myself together. I hope I still find the time to recalculate all the missed tournaments, so that by the end of the year I can compare all three systems used by AlexPank for calculating ratings with his alternative one featuring aging. I intend to calculate a blitz tournament of five rounds in it as follows: I'll sum up each player's ratings over the five rounds. The leader gets 100 points for their figure. For the others, depending on how far their figures lag behind the leader, I'll calculate based on my own formula. That's it. The alternative Rating will do just fine as is. But how will AlexPank calculate the rating for a blitz in one of the three rating systems he manages called the "IronAxe system"? I would suggest doing it as I described above.

Added after 14 minutes

the rating from IronAxe does not differentiate participants with the same days, but DIFFERENT places, which is not only illogical, but also reduces the role of more popular stages (especially the top three) where results are inherently tighter

Yes, I focus ONLY on the primary criterion for determining the winner in a tournament. If it's days - then days, if skeletons - only skeletons, if experience - then experience. Perhaps a person who lags behind the leader in secondary parameters simply didn't have enough time to polish their Walkthrough for the SECONDARY parameter as well. Although they grasped the essence of the Walkthrough according to the PRIMARY parameter just as correctly as the leader did. I believe such an argument has a right to exist and is not devoid of logic. The same person who has time to polish the Walkthrough by both primary and secondary parameters for a given tournament, but it happened that they share a place with someone who lost to them in the secondary parameter and received the same RATING, according to the site administration's design SHOULD receive compensation in GRAILS, as that is exactly where the occupied place SHOULD be taken into account and where the secondary parameter - for example mercury or money - already affects the place. In other words, my rating considers the main tournament parameter and does not consider the place or number of participants. Simply put, the best of the best should receive COMPENSATION in the form of grails for their bells and whistles regarding secondary parameters. Like, collected more resources - get more grails. And the rating will have to be shared with those who didn't lag behind in the main tournament parameter.

Added after 11 minutes

There is only one principle so far: the difference between places should grow as you get closer to the winners.

And I don't understand why that's necessary. Just get as much as you lagged behind the leader in the main parameter and that's it :)

But this condition is unequivocally impossible to fulfill; one will have to pull numbers out of thin air one way or another, because everything comes down to how quickly the rating curve should drop with the place. When that is known, you can simply write some power from 0 to 1 in the oscillator formula instead of the expression n-1/N-1
and get any curvature of the curve. And this is far from the only way. Therefore, to avoid endless arguments, which also won't have objective arguments, I suggest taking an already existing distribution system, using the VDV Forever system as an example.

And how is VDV_forevera's curvature better than the curvature that players will get depending on their lag behind the leader's result in the main parameter? Lagged behind the leader by one day - small curvature. Lagged by 100 days - totally skewed :) specifically.

Added after 5 minutes

Regarding the introduction of this formula, we need to think... but not about the formula, but about the philosophy of our system, i.e., we need to precisely determine and understand what we want from the rating! Who it should reward, and who it should punish. And after that, writing a formula that satisfies our criteria is no longer difficult.. ;)
Respectfully.

Here's what I want - for the rating to reward or punish no one - introduce GRAILS faster for this purpose. Rating in my understanding, especially in Heroes, is a relative value, the reference point of which can only be the result of the leader of a given tournament and the results of other players in relation to them, and ONLY according to the main parameter-criterion for determining the winner (for mercury, place, and number of participants - pile up GRAILS by the bagful :)

Added after 9 minutes

This is a motorcycle racing system, the probability of crashing when you're riding at the limit is quite high. Therefore, the reward for victory should be appropriate. In VDV Forever, the difference is 2 times smaller. Take your pick, I don't want to.

Everyone on motorcycles :) Let's hit the off-road and sloppiness with a road trip :) Joke - kirgudu :)

Logicality and clarity are not the same thing. A logical function is one that satisfies some properties we need. And a clear one... I think it's obvious what I mean. Everything is determined by the mathematical preparation of those around. And I mentioned the power just as an example.
I completely agree, the main question is indeed in the philosophy of the system, but this question is highly subjective.

On one hand, you want to make the rating calculation philosophy clear to everyone, and consequently, you want the rating calculation formulas to be clear to the masses, BUT on the other hand, you're trying to screw some god-knows-what onto the philosophy and rating I proposed :) An oscillator, coefficients + rewards and punishments, a function with power and a gap between places by curvilinear straightness, and throw in motorcycle racing and ping-pong here too :) - what for?
People are already starting to howl from such over-intellectualizing :)
PS. And anyway, who will be responsible for accidents?? This is if someone, following Fireball, decides to conquer 500 with a post limit, but turns out not as strong as Fire. What if some glitch hits someone on the head? :) Joke - kirgudu :)

Added after 16 minutes

1.2. Number of participants. It matters. Moreover, it's directly proportional - the more opponents you managed to bypass, the higher the level of play. A tournament for five players and a tournament for a hundred -

And it seems to me that Radik and Dmitry won't care in the top three whether they bypassed five newbies or a hundred :) Here you are hoping that the more people there are, the MORE LIKELY it is that many strong players will be there - but what if there AREN'T any like them and no one even comes close, then they'll rake in an awesome jackpot for a hundred newbies. Isn't it simpler to give everyone what they deserved :)

1.3. Average strength of players in the tournament. It is fundamentally important - even more so than the number of players. Bypassing the five strongest opponents in a tournament with only ten participants is more significant than beating a hundred conditionally weak players in a tournament where strong ones didn't participate. But for this, one needs to consider the current rating of the player, as well as evaluate the strength of all new participants, which is fundamentally impossible.

Until the simplest rating calculation system builds up a LARGE database of player ratings. Which is what I advocated for at first. I chose what was simpler and didn't intend to complicate it, however, some kind people were found :) and decided to improve the system immediately without having a rating database for players.

Added after 13 minutes

Correct comparison is only possible for those players who participated in all VK tournaments. Otherwise - we're choosing the best football player among hockey players... :)

Agreed. Therefore, if there will be some ONE single player who doesn't miss a single round - they will be the winner, like a self-taught universalist :) However, if there are many players who don't miss a single round - they will fight for that title. At the same time, as AlexPank says - winners in each version will be announced. But then there will be two stages in each version. And there might also be players who take part in only one of the stages and so what? Whose fault is it? Points will surely be summed up over both stages and that's correct!

Added after 5 minutes

Iron had an idea to set limits for maps of different sizes. I don't know if he considers them now or not.
And the second point is indeed very difficult. But it is precisely what gives the objective level of the tournament (even if only champions gathered there, but they lagged behind the minimum completion by 20 days - then the level of the tournament will be low).

The second point is very easy for you mathematicians. Like, the minimum limit for completing ANY map can be taken as one :) Even if it's impossible. And then see who and how approached this limit ;) That is, we introduce a Virtual Relative Leader Cheater who completes any map in ONE day with MAXIMUM secondary parameters :) All other player results are compared with the result of this virtual player and rating is awarded. The hundred always goes to the virtual one :)

Added after 18 minutes


:smile19: Removed :smile19:
Kyle: flood!



Added after 3 minutes
:smile19: Removed :smile19:
Kyle: flood!

Added after 3 minutes

There is no way to prove that one calculation method reflects the strength of a player more accurately than another. Most likely, several calculation systems of equal accuracy will exist. And the simplest one should be chosen...

You speak the truth ;)

Added after 6 minutes

I want to summarize and add something.
Rating system
- map size does not affect the rating

There is also such a concept as map complexity. It affects it. In this tournament where the starting save is the same for everyone - that's one thing. But in the next tournament - the map is easier than the previous one, and a strong player didn't arrive at the tournament and a less strong one on this map shows the same result as the leader in the previous tournament. And how to account for map complexity??? Map size, number and quality of monsters that need to be broken through, etc. After that, you can just hang yourself :)

Added after 10 minutes

I want to summarize and add something.
Rating system
- map size does not affect the rating

Therefore, map size somehow accounts for complexity. At least there, most likely based on practice, there will be more castles and enemies that need to be destroyed and the distance to them is also greater than on smaller maps.

- we ignore the theoretical minimum, there are many problems with it, but little use

Is that so? The theoretical minimum is ONE day :) And there is a use in that.

- accounting for place, result, number of participants and their strength in the system is not a problem.

The result, yes. But at this stage of rating development, you're unlikely to account for strength, as AmberSoler clearly stated. And other parameters - place (read: oil) and number of people (let's say there will be 100 not very strong players) in an OFF tournament - personally, it seems to me that accounting for these parameters will only introduce imbalance. You yourself spoke, for example, about the lack of time for completion according to the main parameter. And here you're also bringing in secondary parameters.

Added after 13 minutes

I cannot answer for others, but personally your system really does not satisfy me. And there are 2 big complaints to it (I hope you will treat the criticism with understanding):
1. The difference in linear ratings for 1st and 2nd places is equal to the difference for 11th and 12th. This not only belittles the role of first places, but also with a large number of participants very quickly splits this difference between them. And yet, to be completely honest, for strong players it makes no difference whether there are 30 or 40 participants. They will sort things out among themselves within the top ten at most.

Oh, Fireball perfectly shows that accounting for the number of participants is not so important. By the way, outsiders also don't give a damn how many people piled up :) And although they are seemingly unnecessary and not in the way, accounting for place and number of people in the tournament can hit them hard. But why?

Added after 12 minutes

What function do you propose for building the curve, if linear does not satisfy you? The function must in any case account for the number of players, even if it is non-linear.

And if nothing of this is taken into account (neither place nor number of players), then the curve or line results by itself depending on the ratio of player results to the leader's result. Especially if an ideal player who we assume completes any map in ONE day is taken as the leader with 100 points.

Then by what criteria will you make a difference between a warm-up tournament and BB? The only criterion, imho - only

Only by the main ones (speed or experience or skeletons or minerals) without considering secondary victory parameters.

the composition of opponents. That's all. What difference does it make to run a hundred meters or 400 meters, if Ben Johnson is racing next to you, for example?.. And again

We know that Ben is a virtual player - so it's worth running with all your might :) Because personally your strength will be measured relative to him. The closer you get to his result - the more chances you have.

returning to our sheep - without knowing the opponent's strength, one cannot calculate the relative significance of your victory, and therefore correctly account for this result in your rating.

We know the strength - Ben is a SUPER-RUNNER. The relative significance of your victory will be visible immediately and most importantly - it will be comparable with other real players.

Added after 2 minutes

What can be proposed? By default, provide each player with a conditional strength parameter equal to five on a ten-point scale (or 50 on a hundred-point - it doesn't matter). Then from game to game this parameter will be adjusted one way or another. Any newcomer receives exactly the average parameter, not zero. Because the average value of the strength spread of any set of players will tend toward the median as the sample size increases. Previously we started from zero, maybe that is where the error lies... One cannot underestimate the true strength of unknown players, thereby unjustifiably lowering the tournament rating... And in every tournament there are a sufficient number of players whose rating is not yet determined...

Furthermore, we calculate the level of relative strength of opponents in this tournament and consider it as a parametric coefficient when calculating the bonus. And no numbers can be taken from thin air! Someone said here that in mathematics and physics many coefficients are taken out of thin air, and nothing happens... :)
I am not aware of such mathematics yet...

No numbers can be taken from thin air says AmberSoler and immediately screws 50 points to the first random player encountered at the tournament :) One cannot underestimate the true strength of unknown players? And how will we know it. He appeared once with 50 points - raised the tournament rating and bailed. And there could be a dozen such people in a tournament. No way - I prefer ZERO :)

Added after 6 minutes

No one is proposing to use a linear rating in its pure form. That goes without saying. It is only a component of the rating calculation, as it reflects the number of tournament participants... This component cannot be abandoned, otherwise in your example second and third place would receive the same rating, which is also not entirely correct...

But tolerable :) Is the argument over a couple of glasses of oil? Compensate the winner with GRAILS :) And for the sake of excitement, leave them equal ratings for the next tournament :)