Ladder season 2 ended

[piranha]

Ladder season 2 ended now and after calculating all the (38) ladder games that passed 15 turns Igor is standing as the winner for all categories.

On the 1on1 ladder he held the #1 spot for a long time with a huge distance to the #2 spot but I think the two Germans Argammon and Razurandal closed the gap a bit the last weeks. There were some crazy streaks there. The Ukranians who were all in the top positions last season ended slightly lower this time.
ChickenChaser who won the last 1vs1 season came in as #4 this time.

On the FFA side Igor took the #1 as well. Stallone who joined warbarons just 3 months ago took the #3 spot followed by KGB. When we closed the last games there a whole bunch of 8 player FFAs on big maps that generated 150 - 240 ladder points for the winner. The K value that is now depending on the number of turns and category (FFA etc) is starting to kick in so big maps like bullrun and FFAs can generate a lot of points.
Garesch who took the second spot on the FFA won 9 out of 10 FFA games which is ver impressive.

Good guys with Igor and Art took the first place on the team ladder. Our fellow countrymen in Scandiania took #2 and Storm Giants the third spot. A few more active teams this season compared to the previous one which is fun to see. We ended up on 5th place with our team.

The best team players was Igor and Art followed by Ahla and TheVic.

The winners have been credited 3000 credits for #1, 2000 for #2 and 1000 for #3 and all the members of the winning teams got 1500, 1000, 500 credits each.

New season is started and will end on the first of January 2013 in 4 months. If you plan to finnish a TEAM FFA game by then you better hurry! .


Sometime during this period we will launch 0.9 which has many nice updates. Our testing of real time games are working better and better. I've spent time on optimizing the amount of data sent to the client with good results. When attacking the amount of data sent to the client is reduced to about 1%-5% compared to now. Opening some popups like the battle list is also reduced by the same numbers. I know that some people said they often get the waiting animation and sometimes reload of client and this optimization should make a big difference for all of you.
The overall feel of 0.9 should be it's much higher quality compared to 0.8.

I've also created some moderator tools so the plan is to get a moderator group together to help us out.

[Moonknight]

Tip of the hat to Igor (and Art)!

Depending on how much time is required, I'd be willing to be a mod...

Is the full calculation for determining winners when games are closed available? I would have thought the number of cities would have been the biggest determination. Trying to figure why KGB won game id 22601.

Will the ladder pool system be ready in the .9 version?

[jetigig]

What exactly would a moderator do? What do you want moderated? And, like Moonknight said, how much time is involved? I would be willing to help.

[LPhillips]

Great season! Thanks so much for putting all that work into the ladder system!
Well played, to all winners and placers in this season!

It would be nice to see the point system for determining auto-winners overhauled a bit. It bears a lot of discussion, and I don't have anything in particular in mind regarding a solution.

Maybe you could cut the number of points awarded by the auto-win? I would think a 50% penalty (or at least no bonus applied) is appropriate. I know that in "Cold-blooded old time" I was going to win, but the auto-win system kicked in and I had to swallow another FFA loss. I'm thinking the games that are automatically determined should count less for both loser and winner. That is, even if the system is perfected and gives obviously correct results.

[piranha]

The calculation is based on this:
15 points per city.
1 point per unit per level
1 point per 100 gold
2 points per building
3 points per hero, 1 per level
(Available in the wiki for ladder)

LP: In the game you played on bullrun KGB had like 250 and you had like 236 points with that calculation, not the exakt numbers but something like that. We did the first 10 game one at a time and looked made sure the result looked ok and that there were no error in our script. I looked at you game and said to snotling that it was so close. I mean there was no obvious winner in the game.

If you have suggestions to further improve the end calculation we can add that for the next ladder season ending.

The moderators are going to do the following.
Together decide which maps should pass from submitted to accepted as a official map, decide which maps should be ladder classed and decide which maps should be in the maps pool that is going to be used with the match making service. We see it as something that should be discussed a bit and reviewed so the mapmaker can receive a mail that explains that you map is accepted but if you wish for it to be a ladder map these things need to be fixed. As a way to improve the ladder map quality.
Get moderator access in the forum so they can keep the forum clean and structured.
They will have access to the translation tools so it would be good if they can keep the english language correct since we make misstakes here and there. We hope we can get some non english speaking users to create translations for their language. Since we have a russian forum section a russian translation would be nice for example.
Perhaps there can be other things later but these are the things we could use help with right now to improve the quality.

[Negern]

The calculation is based on this:
15 points per city.
1 point per unit per level
1 point per 100 gold
2 points per building
3 points per hero, 1 per level
(Available in the wiki for ladder)

If you have suggestions to further improve the end calculation we can add that for the next ladder season ending.

It does indeed seem odd that a lvl10 pala with The Grandfather-artifact is worth three points less than a 0-wall city with crow production and 15 income. I would make heroes 4 + 2 for each level, add points for artifact (1 for lvl1, 3 for lvl2, 5 for lvl3) and make the gaps between units bigger. (1,3,5,7,9).
Income per day should also be counted. Perhaps slightly higher than the total amount of available gold. 1 point per 50 gold per day?

[Igor]

City production, in my opinion, should be counted as well as produced units are. To have Heavy Cavalry as production may be better than have 1 the same produced unit and no production in the city.

[smursh]

Cities could get broken down based on their characteristics. Instead of a flat 15 points, why not something like this
+0-5 points based on the top unit they produce.
points for wall level: +1 for each two points over level 1(max 5)
income: +1 for each 20 gold over 10(max 5)
Base level: 10 point flat value per city.

In this case a city with no production and minimal wall/income is worth 10 only pnts, highly upgraded cities could be worth up to 25. These values might need to be adjusted for balance, but would take into account everything that is invested in a city.

[LPhillips]

LP: In the game you played on bullrun KGB had like 250 and you had like 236 points with that calculation, not the exakt numbers but something like that. We did the first 10 game one at a time and looked made sure the result looked ok and that there were no error in our script. I looked at you game and said to snotling that it was so close. I mean there was no obvious winner in the game.

Probably addressing Moonknight here? If I was that close then the calculation is way, way off, as I never could get anything moving in that game. It's a FFA on Illuria that I was referring to.

This definitely warrants a lot of discussion before decisions can be made. And there are things that will never be appreciable for a computer running stat calculations, like location and unit positioning. That is why I think that the points lost/gained by games terminated for season end should be reduced by some factor. 50% is still my standing proposition. Many FFA games are just getting rolling by turn 15, so awarding hundreds of points for an automatic early termination is a bad thing for the ladder.

The system Piranha has put in place, with each map gaining a "K value" to be applied to the point reward equation, is excellent in normal practice. However, it should be noted that it backfires when combined with this early termination system.
1) Games that would normally take a long time have a high K value, and are worth more points than others.
2) Games that would normally take a long time are least correctly decided by automatic termination, no matter how good the equation becomes. A matchup that takes 30-50 turns to play is not far enough along to judge a winner by turn 15.
So it is obvious that this is a critical flaw in the system. The longer the future of the game, the less likely the equation can accurately determine a winner, and the higher the reward/penalty for the winner/loser.

[LPhillips]

On further thought, a system which ends the games based on a % completed of the average K value for that matchup would be ideal. No hard 15-turn rule (though that could be a minimum), but instead a floating value that is determined by how long the game would normally take. This will remove the previously mentioned critical error. It should also make it more difficult to try to abuse the system in ways discussed in other threads, as players will not know which season a game will count toward. Some people will hate losing that control, but it would actually be very good for the game and the community.