Questions, answers, and resolution of disputes regarding the online rating in HoMM5.

Hello everyone! :) So, I'm posting the revised rules and the formula for calculating ranked duels for 2009. If there are no serious objections, I will post the table with all reports for 2009 on June 1st. Note: All reports starting from post #215 will be used for the calculation, since post #214 by Starkhan was the last one with the rating calculation. If I'm wrong, please correct me, as there are reports from the end of 2008 as well. Actually, the most-most wonderful one kindly agreed to help me correct the old formula, so all respect goes to her (and she's writing her thesis right now, so show even more respect ;) ), and you can throw stones at me :smile08:. Rules (no significant changes): - Players agree on the number of duels in the planned match. - An average of 10 is taken, which is the number of heroes that players eliminate from each other. By agreement, players can eliminate the desired number of heroes. Even if it's two, don't be surprised by a 2-2 or 5-5 score :D - After the elimination, not just one duel is played, but a series of duels (usually 4 or 5 games), naturally, without re-selecting the hero, that is, the used heroes go to the list of the player who played with them, as well as the eliminated ones. - The player who created the game chooses the map on which the game will take place, then the second player chooses a hero (from their non-forbidden ones), then the first player chooses a hero (from their non-forbidden ones), and the game begins. - Dynamic combat is disabled. - The "Boarding!" and "Abyss" arenas are prohibited. Again, if the players agree, I don't see any reason to prohibit these arenas. If you want a pirate boarding, please do :), because a cooperative effort is a voluntary thing :smile01:. - It is mandatory to make a replay of each battle and keep them until you see that the game has been counted in the rating. - Any side, whether winning or losing, writes the report, but with mandatory confirmation from the opponent :smile33:. - In case of a connection break or game crash, by mutual agreement of the parties, the duel: A. is replayed B. victory is awarded to the side with a clear advantage C. the game is not replayed, and a report about it is not written. The selected heroes go to the list of eliminations if this duel is not the last one. - At the end of the year (say, three weeks before the end), the top five ranked players must play three games against each other (each with 5 duels, with 12 eliminations) and determine the strongest. Rating Calculation Formula The initial rating of each participant on January 1, 2009, is 750. New players also start with a rating of 750 by default. The rating is calculated using the formula Rn = Ro + 30 + 0.04 * D in case of victory. Rn = Ro - 15 + 0.04 * D in case of defeat. R new - new rating, R old - old rating, D - the difference in rating, and for the player with the higher rating, it has a negative value. When the difference in rating is more than 250 points, the parameter D is equal to 250. Thus, 0.04 * 250 = 10 or -10. 1. In the case of a weaker player losing (i.e., the difference in rating is more than 250), the stronger player will be given +20 points: Rn = Ro + 30 + 0.04 * (-250) = Ro +20 2. In the case of a weaker player winning against a stronger player, the weaker player will be given +40: Rn = Ro + 30 + 0.04 * 250 = Ro +40 3. When a stronger player loses to a weaker player, the weaker player will have -25 deducted: Rn = Ro - 15 + 0.04 * (-250) = Ro -25 4. And when a weaker player loses to a stronger player, the weaker player will have -5 deducted: Rn = Ro - 15 + 0.04 * 250 = Ro -5 That's all. If the rating difference is less than 250, the calculation is standard (see above). Played duels are counted separately. In strict order, as they were played. Regarding the comments:

I meant to count tournament and regular ratings separately, as they are awarded differently in tournaments. :smile02: Update in the sense of accounting for reports? Then, of course, it can be done once a week :smile14:

Okay. The rating for non-tournament and tournament competitions is kept separately. I agree. Yes, accounting. It can be updated less often, I agree.

How interesting is it "for yourself"? Like, the rest can only eliminate worse heroes? In my opinion, 8-10-12 is the most convenient and good option. Enough already to be angry with everyone, what's the big deal if someone made a mistake :)

What are these old squabbles? :D Stop it ;) And regarding 8-10-12 eliminations, I think it's like this: 10 eliminations +/- 2 is optimal, as reducing the number of eliminated heroes increases the likelihood of the match ending in a draw. That is, a large selection of heroes provides more room for maneuver and there is always a strong counter to the chosen one. Thank you for your help and active participation in the discussion WisTrit, Alex_R, The Most-Most Wonderful, Lord Vampire :smile54:. Additions/Corrections: 1. I received an answer to my request from Lord Vampire - the calculation of reports will start from post 218 of the corresponding topic. All other posts are already included in the results for 2008. 2. alex_nv found an error in formula 1 (when calculating the awarding of points to the "stronger" player, in case of victory over a player with a rating difference of more than 200). Fixed. I don't think this will significantly affect the balance. Well, it's clear that someone carefully checked the pile of formulas. And I wanted to cheat the veterans, you see ;) It didn't work! :D :D Please come to the table, let's analyze everything :) OK! DON'T KICK! TOMORROW MORNING THERE WILL BE A TABLE! I NEED TO RECALCULATE! THANKFULLY, I HAVE A CHECKER! :) YULYA, THANK YOU! THANK YOU FOR REMINDING ME OF THE BASIC MATHEMATICS OF THE 5TH GRADE! :D :D :D