IronAxe
Auto-translated
How do you suggest calculating it? I understand that the system is not perfect, but we need some way to determine a winner. Just by days is clearly not enough. The simplest example: in football points, two teams take the same place. Then the difference between goals scored and conceded is counted, the number of wins, and so on. And no one disputes this.
Added after 9 minutes
But did you check the example? And it's not just about the winner—surely some player with 4 first places doesn't deserve first place in the rating?
By the way, you always claim that the level of play is determined by days, and parameters are just secondary. That the main thing is to grasp the idea. But that's also not true. The one who has better parameters did everything more carefully and was able to focus on parameters, and therefore played better. Moreover, remember the 3rd stage of VK and compare vbn's walkthrough and my finished DmitryA.S. one up to 7 days. They have completely different ideas, and the walkthrough with the sphere is better because the sum of parameters there will be definitely higher. And if it had been done at a tournament, it should have been rated higher.
And regarding secondary parameters, I already gave you the football example above. The score is 4:4, but those who ran in red shorts did it more beautifully and covered more kilometers (meaning they collected more oil), so to hell with them :)
I also wrote that a player might be super strong, but didn't have real time to further polish the walkthrough for secondary parameters. Yes—he'll lose the tournament, lose out in the grails, but he won't lose his Gameplay rating under my system because of this. I have the right?
Added after 31 minutes
But did you check the example?
1. The first player—not me :) who did you copy from—are these real results from a tournament?
Added after 1 hour 10 minutes
PS. As far as I understand, the system with a Virtual Player with 1 day causes no objections specifically for ONE tournament. Moreover, it eliminates the flaw of lacking a strong player with a good result. That is, players who would have followed him won't move up to a higher place and consequently won't take a free rating for that place. I am not talking about the number of players in the tournament at all and don't want to hear about it :)
In the system, as before, the leading and only parameter is DAYS!
However, in reality, the flaw appears when calculating an All-rounder Rating under such a system (I pointed out Fireball's flaw). A small artistic touch (from "art" meaning "bad") and we get a universal system again tied ONLY to days. No games—only sport :) Now there are minimum flaws. In each tournament, the result of each player is compared not even to the leader (it's possible that even the leader didn't find the optimal option), but to the ideal option of the Virtual Player.
----------------------------------------
Calculation of All-rounder Rating according to
the system with a Virtual Player completing
any maps and any stages in ONE day.
player Axe
1st stage - 13 - 769
2nd stage - 13 - 769
3rd stage - 9 - 1111
4th stage - 3 - 3333
5th stage - 17 - 588
Total 55 days - 6570 points
All-rounder Rating of player Axe:
6570/55=119.45
-------------------
player X
1st stage - 12 - 833
2nd stage - 7 - 1428
3rd stage - 7 - 1428
4th stage - 6 - 1666
5th stage - 9 - 1111
Total 41 days - 6466 points
All-rounder Rating of player X:
6466/41=157.71
--------------------------------------
Fireball, now can you find something to nitpick again in this system??? :) The calculation is shown for five tournaments across different versions of the game: Heroes 1,2,3,4,5. But the same can be done if all five tournaments were, say, only for one version, say Heroes-1.
Drawback: All-rounder rating now cannot be calculated if the tournament is not for speed, but say for experience, skeletons, resources. I don't know yet how much experience or skeletons an ideal virtual player should gather and how to bring this to a common denominator with the previous speed tournament. However, both previous systems and mine also suffered from this slightly. Because ratings for speed tournaments and skeleton tournaments were previously a bit absurd when compared to each other; it's just that we always relied on a real leader, but who could claim their walkthrough was ideal.
The same way I see the solution for now—we'll have to calculate however-the-hell much theoretical skeleton collection there can be on a map or how much maximum experience can be obtained and accept this as the ideal option for an ideal virtual player. But it will still be a bit absurd because imho you cannot bring a skeleton collection tournament and a speed tournament to one common denominator. But this trouble was observed before, although we calculated these tournaments using all three systems.
Another trouble—if we are calculating an all-rounder, then only if they played all five stages. Otherwise, I'd play just one 4th version. I'd grab some points there and demand an all-rounder rating of 3333/3=1111 :) And what the hell kind of all-rounder are you if you didn't play at least one of the versions?
Here, the organizers should have their say :) My opinion—if you couldn't handle the all-rounder, fight in your version for a prize place—that's even better and you'll only spend two months on it :)
If anyone has a sensible proposal on this matter—please step forward.
And maybe someone else will find some flaw, but please project it onto your own calculation systems as well and see what comes out of them. We are waiting :)
PS. Actually, to hell with all of it. I won't appear here until I finish the sixth stage :)
Added after 7 hours 3 minutes
How do you suggest calculating it? I understand that the system is not perfect, but we need some way to determine a winner. Just by days is clearly not enough. The simplest example: in football points, two teams take the same place. Then the difference between goals scored and conceded is counted, the number of wins, and so on. And no one disputes this.
And now for the theses (July ones :)
1. Determining the winner by nominations in each of the versions is more or less clear. In principle, any of the three systems Pank uses to calculate the championship rating can be applied there, or even just the VDV system. But how will the All-rounder nomination be determined?
2. By summing the points obtained from each nomination, or by finding the average number of points by dividing the sum of all points by the number of games played, or the method I proposed above: finding the average number of points by dividing the sum of all points by the number of days spent by the player on completing all games? Imho nothing fits, because everyone will play a different number of games and the smartest person would stop at one game where they took first place. He'll beat everyone on average and THIS HAS ALREADY BEEN DISCUSSED—whoever doesn't understand this can go to same garden :)
4. There is a greater chance that a player who played only in their favorite version will get a HIGHER average than a champ player who played all of them, because they might have played some versions for the first time and got hammered there, so the average will be LOWER. With the sum it's also clear—whoever missed more rounds has less chance of winning as an all-rounder.
5. Here we either count the all-rounder only among those who didn't miss a single round. Or if Pank says they will count EVERYONE who played at least 6 rounds, THEN those who played more than SIX rounds must be given the right to choose the best results of 6 rounds from all their played rounds and inform the organizers that for the all-rounder calculation only the rounds specified by them should be taken into account!!!
6. Fireball :) Will you say it's demagoguery again? :) Regarding your statement that all VK stages should be equal—you guessed wrong—that will NOT happen. Example: Player A is strong in 4, player B in 3. At the stage for 4, there were hardly any people. Strong player A took 1st place there, and player B took 2nd place even though they play very poorly in 4, simply because there were hardly any people there. And as we see from Pank's and Amber's systems, PLACE gives a decent bonus (28 points :). What chance does player A have of taking SECOND place in the 3rd version (if they play the third one as crookedly as player B plays the fourth) given that in the third there are people like cockroaches and plenty of pros. Player A will clearly be pushed way further than 2nd place there. And where is the equality of stages for 3 and 4 for players A and B? Mathematicians, hello? This nonsense of inequality will manifest more in Pank's and Amber's systems. In mine, the manifestation of this nonsense will not be as great—only days are taken into account there, and it doesn't matter what place you took or how many people were there. But the ideal remedy against this flaw is the system mentioned above using a virtual player purely in Axe's system, where eventually they calculate the ratio of total points from all stages to the days spent completing all stages. Or not all stages, but only the six best stages for each player. Right?
7. Since not all players play all stages, and the minimum bar is limited by organizers to SIX. I think it's a good idea—to give players who played more than six stages the right to choose their 6 best for the ALL-ROUNDER calculation. Players get at least some options and more people will enter the All-rounder nomination on more equal terms.
Added after 2 hours 21 minutes
:smile19: Removed :smile19:
--Flaw with player rating protection--
Systems that include PLACE and the number of players in their formulas have weak protection against cunning actions, if people are found who decide to use a breach in the defense.
1. Number of players—one can use a clone or several clones of their nicknames. Alternative—several players during the tournament discuss and develop a joint walkthrough plan, but then each performs the walkthrough independently. Proving a violation is IMPOSSIBLE. Amber! Confirm the correctness of this statement, otherwise I'll find YOUR post where you talk about this myself!
2. In such a layout, we have an example.
Player A - 2 days - 1st place.
Player D - 4 days - seems to be aiming for 2nd place.
But then two groups of players B and C appear. As a result, we have:
Player A - 2 days - 1st place
Player B1 - 3 days - 2nd place
Player B2 - 3 days - 3rd place
Player B3 - 3 days - 4th place
Player C1 - 4 days - 5th place
Player C2 - 4 days - 6th place
Player D - 4 days - 7th place
(pushed to the back, here even secondary parameters for determining the winner helped)
We assume that group B developed walkthroughs jointly and group C also developed a walkthrough jointly, but separately from group B. But such a thing can happen. Amber himself recently bothered me with a stupid example from the Landless Tournament on HP. He even said that all options, even theoretical ones, must be considered to check the system's viability :) Well damn, now consider YOUR SYSTEM with an oscillator :)
3. Player D is specifically pushed down in rating DUE TO the accounting of the PLACE taken, because a lot of cunning people appeared (accounting for the number of players in the tournament) + accounting for secondary parameters for determining the winner. If the proponents of rating systems that account for place, number of people, and secondary parameters for determining the winner DO NOT FIND PROTECTION AGAINST THIS FLAW, then they can go to the bathhouse :)
Your demagoguery regarding honesty, honor, and other virtues DOES NOT INTEREST me. In this system, a player cannot feel protected from artificial lowering of their rating because it will be impossible to prove. Now project this situation onto Axe's rating. Nothing for all the cunning ones! The Axe has iron protection :) Player D doesn't care about any bullets in such a system. Only DAYS are tied there and he gets his honest rating corresponding to his result in the tournament, which will correspond to THIRD place, not SEVENTH as per the mathematicians' system :)
4. Conclusion: The more parameters a Heroes rating calculation system takes into account, the more flaws related both to the work of the formula itself, protection, and many other things :) Study your materials, mathematicians :)
--------------------------------------------
PS. In the discussion of the last tournament, where there was a small fuss with the Boots clan, I wrote a post saying that as an accountant, I don't give a damn whether they developed the walkthrough jointly or separately, because the rating of players whom they pushed and who took places below them WILL NOT SUFFER. The pushed players will get the rating that CORRESPONDS TO THEIR RESULT IN DAYS. And Amber wrote something like "let the accountants think for themselves what is what," but first HE himself shoved a damn oscillator at them :)
PSS. Amber, good luck coming up with a coefficient that eliminates this flaw, which in my system is already absent by itself! As you can see, not everything can be accounted for by mathematics :) I state once again that a rating calculation system without accounting for place, number of people, secondary parameters, and bonuses is MORE STABLE and RELIABLE. Its price-quality ratio is much higher than any other :)