Five Tribes Five Tribes

Forums

Search
Forums » Memoir '44 Online - English » Discussion about the Skill system
Show: Today's Posts 
  
AuthorTopic
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Discussion about the Skill system Mon, 13 December 2010 21:46
Dear M44 Online Officers:

There were many comments and suggestions lately about the Skill system.

We would like to develop the various ideas and see what comes out of the discussion. We are aware that the current system is not perfect, even though it has some virtues that might not be obvious to everyone.

For this reason, we wrote a memo that describes the current system in details at http://cdn1.daysofwonder.com/memoir44-online/en/img/M44-ELO- Ranking.pdf
Please start by reading it. It explains the various issues and how we tried to address them.

It is possible that we are trying to kill too many birds with one single stone, and that we ended up with a mathematical monster that is just too complex to be understood and accepted. That's why we would be happy to explore alternate ideas with you.

One thing should be very clear to everyone: if we cannot find a system that is satisfying to both all M44 players and us, we will pull it out. Likewise, if the Skill system becomes an obsession, we will pull it out.

The main reason why we kept an ELO system was that it was very successful with Ticket to Ride Online. But may-be a bit too successful! Some Ticket to Ride Online players are so obsessed by their ELO ranking that they play only with the best players, behave in a rude manner with others, etc. We would rather favor a more casual, honor-based style of playing for Memoir '44 Online, which in our opinion is more consistent with the game. That's why in our minds, the Officer Ranks and Achievements are a better representation of a player's experience.

At the same time, we are aware of the demand from several competitive players. So we look forward to your ideas!

The Memoir '44 High Command at Days of Wonder Headquarters
      
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Possible Alternate Ideas Mon, 13 December 2010 21:51
Among possible alternate ideas to explore for a Skill Leaderboard:

- Reset everyone's Skill value on the first of each month, and keep a snapshot of the Leaderboard of the previous months to create monthly Leaderboards: "Best officer of November 2010"

- Keep the Skill system only for (future) online tournaments - in a format that has yet to be defined.

- Disable the Skill in all games, except if players specifically activate it (note: this introduces a bit more complexity in the user interface of the Briefing Room. And there would probably a need for an additional filter in the Scenario List).

Yann

[Updated on: Mon, 13 December 2010 21:52]

      
philmcd
Member
Major

User Pages
Posts: 39
Registered:
November 2007
Re:Possible Alternate Ideas Mon, 13 December 2010 22:08
Well, we already had a thread, but as you've started another one, I'll post my suggestion here too.

If the ranking system can't be made to MEAN something... uh meaningful then it might as well be scrapped.

Personally I would favour keeping it, but ditching the current system of calculation.

Might sound complcated (but bear in mind the computer does the brain work).....

If you win, you score the reciprocal of the percentage that your side normally wins the scenario.

Lets say that a scenario is won 24% of the time by Axis and 76% of the time by the Allies.

An Axis Win would gain you 7.6 points or a loss would lose you 2.4 points

An Allied win would gain you 2.4 points or a loss would lose you 7.6 points

It can therefore be seen that if (as is hoped) opponents play both sides of a scenario, You can win exactly 10 points if you win with both sides, and lose exactly 10 points if you lose with both sides. A win and a loss would result in either a small gain or a small loss of points. Removing the wild swings currently in evidence.

This would be a far better system than the current one because the difficulty of the scenario is the predominant factor, and better players would tend to rise to the top.

A monthly and annual table would be interesting for many players, and good for business for DOW.
      
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Re:Discussion about the Skill system Mon, 13 December 2010 22:11
Yes philmcd, I saw your original message on this matter - and found your idea interesting. I created a new thread because I wanted it to be sticky for a while and give the opportunity for everyone to download the attached PDF.

This is typically the kind of feed-back we are looking for. Smile

Yann
      
gheintze
Senior Member
Brigadier General

User Pages
Posts: 949
Registered:
August 2004
Re:Discussion about the Skill system Mon, 13 December 2010 22:23
Hi Yann and the DoW crew,

First of all, I just want to say that I'm very impressed with how responsive DoW is to the comments and feedback on the forums. It was great that you changed the gold ingot system when that was the hot topic and now you are addressing the ranking system, which seems to be dominating the forums these days. This is why DoW is one of the better game companies and one I am proud to support. I only wish some other companies (you know who you are) would follow your example...

As far as the ranking system, given all the constraints and variables of M'44, I think that it works pretty well overall. However, I do think that the amount of medals you win by needs to be a little more important in the formula. I'm not a mathematician, so I can't understand everything in the PDF, but my empirical observations are that whether you win or lose and the strength of your opponent are the most important factors.

I also think that if you want to develop a friendly atmosphere with M'44 (contrary to that observed with TTR) then you need to deemphasize the points gained or lost due to the difference in skill levels. That will not punish experienced players for helping newbies out. For example, I played Arnhem as the axis a last week with someone who didn't even know the basic rules. I made moves to illustrate how tanks worked, etc.. that may not have been in my best interest in order to teach. This should not be punished too much in the rankings.

I think that if you just make the whole system more static (less punishment for losses, less gain for wins) people will be happier. It will help to mitigate poor dice rolls, bad cards, etc... but over time the cream will rise to the top. This will also make for more friendly competition if there isn't so much at stake.

Also, I personally take great pride in the achievements and promotions. I think it is a strategic challenge to try to get some of the medals (killing 5 units with elite infantry, for example) and adds an interesting level of complexity to the game.

In order to encourage this, you are going to need to keep adding new medals to keep the interest level high.

Promotions are great because they require a long term commitment to playing well. Excellent -- just don't make the requirements so obscure that we can't figure out how to achieve the goal (both promotions and achievements).

But most of all -- keep bringing out new scenarios. If i remember correctly, there are about 14 more scenarios that only use the base game rules before the terrain pack was released. I can't wait to play them! You will need to work hard to keep it fresh and exciting... but having the awards will help with that.

Keep up the great work! I'm really enjoying M'44 and have started reading more about the history of WWII. Thank you for the many hours of enjoyment.

Geoff
Captain in the 95th Infantry Division, US Army

[Updated on: Mon, 13 December 2010 22:29]

      
Randwulf
DoW Content Provider
Major

User Pages
Posts: 1362
Registered:
March 2005
Re:Discussion about the Skill system Tue, 14 December 2010 00:05
WoW... cool... looked over...

Math fumes... cough cough...

okay... here is the problem as I see it.

Your base points per game are set to high!!!

Reduce the base points by half or even two thirds and the rest of the math still works, but you don't get the wild point spreads.


And I know it will be a pain but try to round off the outperform
numbers.

Some of the scenarios in order to outperform you must win it.
In a 6 to 5 game it hurts the lower side when the outperform number is 5.2 or higher??? even the games that are an even 5 medals you don't outperform unless you get MORE than that number. It needs to be equal to that number.

But that is just how I see it.

Your mileage may vary....

      
Sgt Storm
Senior Member
Lieutenant

User Pages
Posts: 887
Registered:
December 2006
Re:Discussion about the Skill system Tue, 14 December 2010 00:10
I would prefer to see skill level adjustments made based not on winning but on outperforming. The scenarios are unbalanced, apparently by design. I should not have to play the underdog x number of times to get rewarded for being courteous enough to take the underdog side when I host a game.

If you cannot base the skill level on average performance rather than outright win/loss then there are other suggestions:

1. Remove the skill system since it is not appropriate in unbalanced games (I would prefer it not be there at all - why not encourage play and not the angst and in-fighting that comes to some players from the skill level ranking).

2. Re-implement the concept of a game in the online system as a pair of games on the scenario with each side switching. Only apply skill level to those games played where the sides swap and play the scenario twice, with an option to play one side without skill level being affected.

BTW, when I originally brought up the issue of balance in scenarios in the board game, maybe a year ago, I was "reminded" by many that you are supposed to play the scenario twice, once on either side. I actually still believe the scenarios can be balanced with a little effort without sacrificing historical integrity, but apparently very few others do. So, implement the online game as it is supposed to be played and only apply skill level rankings in those cases.

But, I think it would be simpler to base skill level on over/under performance and be done with it (or remove it altogether - my preference).

Under no circumstances should it be based on win/loss, IMO.
      
Talespinner
DoW Content Provider
Rikugun Taii

User Pages
Posts: 119
Registered:
March 2006
Re:Discussion about the Skill system Tue, 14 December 2010 03:12
Yann wrote on Mon, 13 December 2010 13:46

Some Ticket to Ride Online players are so obsessed by their ELO ranking that they play only with the best players, behave in a rude manner with others, etc.


And this is precisely why I do not play Ticket to Ride Online, except in games with friends when I have been specifically invited.

I say drop the whole system altogether and let people play the game and have fun. Use Ranks and Achievements to get an idea of skill level, but do not let a few "competitive" players ruin the game for everyone else. If you want me to break out the list of games that failed in the past 10 years because of "several competitive players" who demanded and whined until their "way" was "law" I can this post turns into a Russian Epic...and no one wants that, we would all rather just roll dice and blow things up.
      
Nightrain
Senior Member
Oberstleutnant

User Pages
Posts: 424
Registered:
October 2008
Re:Discussion about the Skill system Tue, 14 December 2010 04:02
I came up with an idea, an advance calculation of phil's previous idea. This math will take into account of your outperformances and also the accomplishment of winning by large margin of medals

to simplify thing, don't take into account the difference between ranks and the good thing about it is you're welcome to play any players with the same odds

i think the base point of 10 probably too low, let's try it with 20, but we can try adjusting it to 12 or 15 later

add another variable of medal achievement with this math :
if medal gained > 0 then
( medal gained / total medal allowed ) x ( scenario odds x 20 ) = your point
else ( scenario odds x 10 ) = your point

for example :
losing as Allies (76%) with 5-6 score, still very close
( 5 / 6 ) x (76% x 20) = lost 12 points (rounded down)

losing as Axis (24%) with an outperformance 5-6 score
( 5 / 6 ) x (24% x 20) = lost 3 points

This way, the outperformance are calculated with the score,
so losing with 0 medals will still punish you with the lowest score of the calculation

Another advance calculation for winner is to also take account how many medal lost, so if you win in a scenario by 6-0, you'll get an additional point

math is :
bonus point of 10 - (medal lost / total medal allowed )

for example :
winning 6-1 will give you a bonus point of (10 - (1/6)) = 8 (rounded down)
winning 6-4 will give you a bonus point of (10 - (4/6)) = 3

if you win 6-1 as Allies (76%), you'll get a total point of 4 + 8 = 12 points
if you win 6-4 as Allies (76%), you'll get 4 + 3 = only 7 points

another big score is to win as Axis (24%) by a good margin, let's say 6-3 you'll get 15 + 5 = 20 points

what do you think ?
      
rasmussen81
DoW Content Provider
Designer's Oath

User Pages
Posts: 7091
Registered:
July 2007
Re:Possible Alternate Ideas Tue, 14 December 2010 05:24
Thanks for bringing an official ear to this discussion, Yann, and thanks for your willingness to listen to your fans! As with any discussion, there will be different opinions, different points of view, and different priorities. I think it's important to recognize that the final decision in this area is going to leave some people happy, others unhappy, and some people indifferent but that's going to happen no matter what we or DoW does...

Yann wrote on Tue, 14 December 2010 00:51

Among possible alternate ideas to explore for a Skill Leaderboard:

- Reset everyone's Skill value on the first of each month, and keep a snapshot of the Leaderboard of the previous months to create monthly Leaderboards: "Best officer of November 2010"


This idea appeals to me because it would be fun to let people battle for the top spot each month and it eliminates the current feeling that nobody can catch the current leaders. Every month the table would be cleared to leave room for a new "Top General"! DoW could even create a unique badge or Achievement that could be proudly displayed (like our current Military Rank) that would show off the month and year that we were Top General!

I seems that maybe this idea should be partnered with other plans for changing the Skill system because I like this idea no matter what other changes are made to the Skill formula.

Yann wrote on Tue, 14 December 2010 00:51

- Keep the Skill system only for (future) online tournaments - in a format that has yet to be defined.


I think this is a fine idea and the prospect of online tournaments organized by DoW is very exciting! This plan appeals to me because if I thrive on the competition side of things I can participate in every Tournament that is organized, but if I don't like having to play with the skill system I can just enjoy the game without any skill system.

Yann wrote on Tue, 14 December 2010 00:51

- Disable the Skill in all games, except if players specifically activate it (note: this introduces a bit more complexity in the user interface of the Briefing Room. And there would probably a need for an additional filter in the Scenario List).


I can see the appeal of this idea because people could then choose on a battle-by-battle basis if they want to play for points or just have a fun game. It would be more complex for new players but I think people could figure it out pretty quickly. I wonder if people who really enjoy the competition would like this idea because it might mean that there could be a limited number of players who want to compete. This could mean that their pool of potential 'competitive' players would shrink considerably and they might find it hard to find opponents. But maybe I'm underestimating the number of players who really like the competitive side of the game...

So after all of that, I realize that I haven't helped move this discussion forward very far! Laughing My vote would be for a modified skill system that focuses primarily on the difficulty of the battle (instead of the skill of the opponent) and the implementation of the Monthly Best General idea. In addition, there could be periodic tournaments that use the same skill system...

[Updated on: Tue, 14 December 2010 05:25]

      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Possible Alternate Ideas Tue, 14 December 2010 06:21
Hi, Yann --

I echo everybody's appreciation for how open to feedback you and the rest of DOW have been, about this and so many other issues here. You folks are exemplary in dealing with your customers.

I have printed the PDF, and will absorb it so I can offer a thoughtful response. Thank you for providing the details. And for everything else you do.
      
henryblake
Junior Member
Major

User Pages
Posts: 14
Registered:
June 2005
Re:Discussion about the Skill system Tue, 14 December 2010 06:54
Whatever the relative merits or drawbacks of the current ranking, I think that most concerns could be resolved without tinkering too much with the formula, but by instead adding an option to the game setup that forces a two round game. I think most people who would care about ranking generally like to play both sides anyway, so this would kill two birds with one stone.

Single round games would still be ranked, but since full games would be ranked based on the total score of two rounds, the result would be worth twice as much as an individual round. I doubt single rounds would affect rankings that much because I think higher ranked players will likely play full games instead of individual rounds unless they are short on time, in which case more power to them as they try to move up the beach on Omaha. Ranking by game instead of round should smooth out much of the score variability that players keep complaining about.

I try to play only full games (ie, both rounds). This is the way the game was designed to be balanced, and this is what I find fun. I am finding that most of the people I am playing (and I play whoever asks) are also asking to play both sides (so far I have only had one person ditch the second round after he lost the first). For people who plan to play both rounds, the current after action scramble to reinvite is kludgy. It would be better if it were automatic and forced (I don't want a rematch button, I want it integrated into the ranking. So the *game* is ranked, not the round).

Instead of a choice between axis or allies when setting up the game, the choice would be axis, allies, or both. The computer would randomly assign who plays which side first. This should make it clear to the players in the scenario lobby that they are committing to a two-round game.

Some players won't keep their commitment, and some players will be forced by life intervening to end prematurely. Such is life. Abandoned games would be finished by Johnny, and players would still be responsible for their scores.

I don't support the idea of adding the option of non-ranked rounds because it would add complexity to the pre-game dance. It is nice that some players are willing to guide newbies through a scenario, and if they don't care about ranking, great. But since I am paying for each game I play I want to play someone who is already conversant in the rules. New players are welcome to familiarize themselves with the system using Johnny, or are welcome to play me at full strength. War is hell.

Hopefully an observe mode will be added soon that will help new players learn by watching (I am sure this is on the to-do list and is only a matter of time). It would also be helpful for newcomers if there were more in-game help about how a player can use a card as they play it, but these are topics for another thread.
      
Phread
Senior Member
Stiff Upper Lip

User Pages
Posts: 1778
Registered:
December 2008
Re:Discussion about the Skill system Tue, 14 December 2010 07:03
I don't have issues with the current skill system - it is not why I play. I have striven to gain rank and get all available achievements - these are (probably) more important to me.

Clearly some players are focused on the skill points while others aren't.

I do like to look at the rating of the player I am playing against, so I have some idea of what to expect - but it doesn't make me reject an invitation to play.

I think the current system isn't too bad. In the last week I've been between about 1575 and 1650. I've chosen to play against new players and lost as the favoured side and lost a lot of points - that's the way it goes. I've played players ranked higher than me, got good cards and dice and won as the underdog - that's the way it goes. You win some you lose some.

I would be disappointed with some of the suggestions I've seen. I think the skill of the players should be a consideration, how well you played (better or worse than average) should be a factor as should the favoured or underdog status.

I'd be disappointed if the skill ranking was to disappear. Perhaps we could have a cumulative score as well as this month's score, and while we are at it a rank for the year. Computers can do this easily.

BTW Thanks for discussing this DoW, thumbs up for being prepared to start this thread.

Phread - offering his 2 cents worth.
      
gheintze
Senior Member
Brigadier General

User Pages
Posts: 949
Registered:
August 2004
Re:Discussion about the Skill system Tue, 14 December 2010 07:44
stephenwallace wrote on Tue, 14 December 2010 00:54


Instead of a choice between axis or allies when setting up the game, the choice would be axis, allies, or both. The computer would randomly assign who plays which side first. This should make it clear to the players in the scenario lobby that they are committing to a two-round game.


I really like this idea. If it would be able to keep track of cumulative medals and figures to see who won after two battles, that would be spectacular.

Geoff
      
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Re:Discussion about the Skill system Tue, 14 December 2010 07:44
To everyone:

Thank you for your various ideas, this is great. I will compile them, and answer individually to some of them in the next few days.

Keep in mind that we might scrap the whole Skill thing, so don't feel bad if you contributed and that nothing comes out of it in the end.

Again, this is a complicated (and an emotional Razz) matter.

Yann

[Updated on: Tue, 14 December 2010 07:46]

      
Mighty Jim 83
Senior Member
Starshiy Leytenant

User Pages
Posts: 333
Registered:
August 2009
Re:Discussion about the Skill system Tue, 14 December 2010 10:12
Am going to have a bit of a chew over the PDF before making any detailed suggestions on skill, but wanted to make a couple of comments first.

1.) To echo the early comments. Much kudos to DoW for the level of genuine response and attention to comments posted on these forums. You can never please everybody, but at least no-one ought to feel that their view has not been listened to.

2.) In defence of one-way games. As a rule I would always try to play both ways, but that's sometimes just not possible. I've played some scenarios both ways in less than half an hour, and on other occassions, it's taken me 45 minutes just to play one way. That might be down to the scenario a little, but often it's the playing style of the opponent, and the cards/dice we both have. If I've only got an hour or so to spend online and the first game suddenly takes most of that time, I'm not going to start the re-match.
In short: by all means, encourage people to play both ways, but don't produce a skill system that depends on it.
      
Nightrain
Senior Member
Oberstleutnant

User Pages
Posts: 424
Registered:
October 2008
Re:Discussion about the Skill system Tue, 14 December 2010 10:57
Mighty Jim 83 wrote on Tue, 14 December 2010 16:12

Am going to have a bit of a chew over the PDF before making any detailed suggestions on skill, but wanted to make a couple of comments first.

1.) To echo the early comments. Much kudos to DoW for the level of genuine response and attention to comments posted on these forums. You can never please everybody, but at least no-one ought to feel that their view has not been listened to.

2.) In defence of one-way games. As a rule I would always try to play both ways, but that's sometimes just not possible. I've played some scenarios both ways in less than half an hour, and on other occassions, it's taken me 45 minutes just to play one way. That might be down to the scenario a little, but often it's the playing style of the opponent, and the cards/dice we both have. If I've only got an hour or so to spend online and the first game suddenly takes most of that time, I'm not going to start the re-match.
In short: by all means, encourage people to play both ways, but don't produce a skill system that depends on it.


i agree with Jim
it's not that we don't want to play both sides, but sometimes real life prevented us from doing so. Skill, if later still be used, should be capable of calculating a single game as fair as possible Smile
      
rasmussen81
DoW Content Provider
Designer's Oath

User Pages
Posts: 7091
Registered:
July 2007
Re:Discussion about the Skill system Tue, 14 December 2010 11:11
I know that some people will be very disappointed to see the Skill system disappear completely. I even talked to one player (who really enjoys working his way up the Skill ranking) and he said that if the Skill system is taken away he doesn't know if he'll keep playing Memoir '44 Online. Confused

I can understand that some players think it's fun to work toward a higher Skill rating, but we should not let that be the only reason we play this game!! When I bust out the board game and set it up on the dining room table, I'm not playing for some big rank or to move myself up a skill ladder...I'm playing to have fun with a game that I love to play! We can't lose sight of this. Smile

Now if I went to a convention and entered a tournament, I am trying to fight to the top but DoW can still run tournaments without a larger "Skill Rank" system in place.

The one thing we will lose if DoW removed the Skill system completely is the information it provides us about our opponent. Right now, even though many people don't like the system, we all use it to gauge who we're up against. The skill number isn't very prominent (being in small text when you click on someone's name in the lobby) but I do still use it.

If the Skill number is removed, I think it should be replaced with some other information that could help us tell if our opponent has played much Memoir '44 Online. The Rank, which is very prominent, gives us some good information but it only tells us that they met certain requirements...not if they are any good at the game. The trick is, I don't know what you would replace the Skill number with.
      
Pike Bishop
Junior Member
Arnhem Victory

User Pages
Posts: 16
Registered:
January 2005
Re:Discussion about the Skill system Tue, 14 December 2010 12:51
Yann wrote on Mon, 13 December 2010 20:46

Dear M44 Online Officers:

One thing should be very clear to everyone: if we cannot find a system that is satisfying to both all M44 players and us, we will pull it out. Likewise, if the Skill system becomes an obsession, we will pull it out.




Although I personally enjoy working towards achievements and promotions more, I would like to see a skill system of some kind. However I have a feeling that no matter what is implemented it is going to be incredibly difficult to a) satisfy everyone and b) not become an obsession for some.

Having said that, I would like a solution if possible and giving it some thought, I would also prefer a system designed to incorporate the accepted offline scoring a little more - which is the total medal count over two games of the same scenario.

A few people have mentioned it's not possibele to always swap sides and play again immediately, which is completely understandable. However, could the system track all the games played, and only when a player has played both sides of the same scenario, even if that happens to be months apart against different opposition, is the total medal count over both those games used in the calculation of skill?

Pros - It is more reflective of the accepted method of scoring
It will encourage (force?) players to play both sides at some point as nothing counts towards your rank until both sides have been played

Cons - playing against opponents of different ability will probably still be a source of contention
Difficulty in getting opponents to play in the required scenarios
Will it be too much information to store effectively (I'm not technically minded I'm afraid, so have no idea of the demands this will create.

Another thing to consider would be a way to notify players which games still need to be played to count in towards skill ratings

[Updated on: Tue, 14 December 2010 12:54]

      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Discussion about the Skill system Tue, 14 December 2010 14:32
That document describing the Skill/Rating/Ranking system is, indeed, very informative. Here are my current thoughts, after reading it and doing a bit of analysis.

1. The basic concept of the winner gaining X points and the loser losing the same number of points makes sense to me.

2. I also agree that it makes sense for the payoffs to be proportioned according to the historical winning percentages for the scenario, so that, on average, the expected payoff would be 0. (In Omaha, which splits approximately 15-85, if the payoff were 50-50 or even 30-70, playing Allies would be a guaranteed way for virtually all players to lose rating points.) When new scenarios are introduced, I'm not sure how they will be scored, but I'm sure there's a standard method for that.

3. I believe that 1-way games should not be discouraged in any way. They are an important part of Memoir. (A) Memoir's strength comes partly from the fact that it can be played reasonably quickly. Asking people to double their time commitment, especially in an online environment where real world events can pull them away from the computer and Honor can be lost, is troubling. (B) It's not fair to ask a weak player to play 2-sided matches in unbalanced scenarios; after I have a razor-close, fun game at Omaha against a newbie, it's not fair to ask him to take Allies and lose 6-1. Better to move to another unbalanced scenario and give him the easier side again.

4. The "medals" component of the games Point Value should be larger. An Allied 5-4 win at Operation Cobra is +16.31, while an Allied loss is -15.05; however, that 31-point rating swing can easily be the result of 1 die roll. I think that is the problem. Players should gain relatively few points for 1-medal wins, and many points for big wins. (And when we start seeing scenarios with many-medal objectives, this should be based on a percentage of the objective, not the absolute number of medals difference in the score.) I note that the French Federation tournaments are scored on total medals, including medals gained in losses, not on the players' won-lost records.

5. The "bonus" curve per incremental medal should be steeper. 5-0 and 5-1 wins are relatively hard to achieve, and should be rewarded. Rather than increasing at .10 per additional medal, use some appropriate exponent. Try it a few different ways in the test environment, and see what produces results that "feel" right.

There are currently 778 AARs for Operation Cobra where the winner is listed with 5 medals. (I had to discard several games with inadmissable scores.) Among Allied wins, only 13% were 5-1, and 5% were 5-0. Among Axis wins, 14% were 5-1, and 3% were 5-0.

6. The adjustment for the difference in players' strengths may need to be increased. (I'm not sure yet. It depends on how much spread there is in the ratings.) This factor should be re-evaluated after the medal factors have been adjusted. Example: Last night, I lost Op Cobra 4-5 to a player rated approximately 100 points higher than me, and the rating impact was 13 points, instead of the "standard" 15.05, so the strength adjustment seems to be small. In an "even" scenario, a strong player may beat a weak player 75% or 80% of the time from either side, while two strong players may split it 50-50.

So you may want to scale the point value or a win or loss according to how far apart the "strong" and "weak" ratings will become. Unfortunately, there is no perfect answer here. There will be ways for a player to "game" the system and boost his rating, no matter what formula you use. That's a simple consequence of the fact that any player can play any number of games against any opponent. And in the grand scheme of things, it's okay.

[Updated on: Tue, 14 December 2010 14:37]

      
Randwulf
DoW Content Provider
Major

User Pages
Posts: 1362
Registered:
March 2005
Re:Discussion about the Skill system Tue, 14 December 2010 15:54
Regardless of the rank system being kept or scrapped, I would like to see a games played and % won % outperformed in the last 30 days added to our profile. Start at 50% base, if you have not played in a month your base will start over. Track for only the last 30 days, You can have a career total in your service record and your profile page.

I think that would be a better indicator of a persons skill than any type of point ranking system.

If I see a guy that has played 10 games and has won 7 out of 10 or 70% and has outperformed 100% he will be a tougher opponent than the guy that's played 100 games with a win of 60% and an outperform of 75%

But if you find someone that is at a 20% win and 20% outperform, would you want to play him??? would he want to play you??? chances are he would get whooped.

Even with that, I would like to see the % win outperform in the service record.

[Updated on: Tue, 14 December 2010 16:05]

      
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Re:Discussion about the Skill system Tue, 14 December 2010 15:59
Thanks sam1812 for your excellent comments. All your points make a lot of sense! Very Happy Maybe it is just a matter of tweaking the values and putting more inertia in the system?

Still, the non symmetrical property of the ELO system bothers me.

Obviously, I need to think about all this some more... Rolling Eyes

Yann
      
Sgt Storm
Senior Member
Lieutenant

User Pages
Posts: 887
Registered:
December 2006
Re:Discussion about the Skill system Tue, 14 December 2010 18:08
Talespinner wrote on Mon, 13 December 2010 21:12

Yann wrote on Mon, 13 December 2010 13:46

Some Ticket to Ride Online players are so obsessed by their ELO ranking that they play only with the best players, behave in a rude manner with others, etc.


And this is precisely why I do not play Ticket to Ride Online, except in games with friends when I have been specifically invited.

I say drop the whole system altogether and let people play the game and have fun. Use Ranks and Achievements to get an idea of skill level, but do not let a few "competitive" players ruin the game for everyone else. If you want me to break out the list of games that failed in the past 10 years because of "several competitive players" who demanded and whined until their "way" was "law" I can this post turns into a Russian Epic...and no one wants that, we would all rather just roll dice and blow things up.



I don't play TTR online for exactly the same reasons. I don't want this to happen to M44 online.
      
Scragnoth
Senior Member
D-Day Recon Team - Silver

User Pages
Posts: 194
Registered:
October 2005
Re:Discussion about the Skill system Tue, 14 December 2010 18:31
Recently I bought the TTR boardgame and inspired by that had a look at the online game. I didn't play but just participated.
The languague in that game and conversation let me pass on the game online all together.
So indeed, don't do this to M44.

For me the skill level is not of interest but the achievements make it really interesting. I hope there will be enough to discover on that end. Also in the future.
      
Brummbar44
DoW Content Provider
Artillery Specialist

User Pages
Posts: 1129
Registered:
June 2004
Re:Discussion about the Skill system Tue, 14 December 2010 20:50
sam1812 wrote on Tue, 14 December 2010 05:32


5. The "bonus" curve per incremental medal should be steeper. 5-0 and 5-1 wins are relatively hard to achieve, and should be rewarded. Rather than increasing at .10 per additional medal, use some appropriate exponent. Try it a few different ways in the test environment, and see what produces results that "feel" right.

There are currently 778 AARs for Operation Cobra where the winner is listed with 5 medals. (I had to discard several games with inadmissable scores.) Among Allied wins, only 13% were 5-1, and 5% were 5-0. Among Axis wins, 14% were 5-1, and 3% were 5-0.



Sam makes some great points but I have to disagree here...to a certain degree.

I've been blown out on more than a few games because the dice simply fail. So a 6-1 loss with a hit percentage of 30% to the winners say 50% should earn them more points?! The game wasn't truly reflective of my abilities, just poor luck...should I additionally suffer because of that? It's frustrating enough to know that the dice are sour, knowing I'm going to lose more points as a result would be a turn off.

I would think that if the ranking system is to survive, the rolling percentage should be taken into account.
      
Valeroth
Junior Member
First Lieutenant

Posts: 7
Registered:
August 2007
Re:Discussion about the Skill system Tue, 14 December 2010 21:00
To start, I like sams post quite a bit.

Some of my own thoughts:

1. The medal spread calculation should be geometric not linear (as I think Sam suggested).

2. Outperforming the average should have a much bigger impact for the loser, without impacting the winner.

In a game like Omaha Beach, If lose 6-5 maybe I shouldn't lose any points (still not gain any either). Still if I do lose points it should be alot less.

3. Personally, I only care about the skill system once it reaches the point where people won't play a game with me because my skill is to high or to low.

Ticket to Ride is one of my favorite games, and I like playing it online, but I do get a little bummed when I see so many games that only offered for players of a certain skill level or higher.

My personal preference would be to scrap the skill system, but instead setup a series of weekly and monthly tournaments. I do really like the achievement and rank system. In all the tournaments I've played in it was the total medals gained that counted, not the total wins.
      
Axelb9
Junior Member
Major

User Pages
Posts: 25
Registered:
November 2010
Re:Discussion about the Skill system Tue, 14 December 2010 22:46
Although I have voiced my ideas in several earlier threads about the subject matter having read all the opinions here my thoughts are the following:

Create a toggle for a player to set whether he wants to play:

A / a short just for fun single scenario that does not lose or gain skill at all

B/ a full game playing both sides that adds or loses skill based on total medals won during the two games

This should satisfy most everybody. If one has not enough time he/she could still go for a short game whereas if ones is able to devote 45-50 minutes for a full game then he/she would be able to advertise his willingness and find likeminded partners much easier tha today. If one is only ranked when playing a full game based on total medals scored then the current ranking system or any other ranking system is just as OK as the chances have been evened out.

I also strongly support the idea of organized tournaments (based on full games) with permanent achivements won for winning one.

Alex
      
philmcd
Member
Major

User Pages
Posts: 39
Registered:
November 2007
Re:Discussion about the Skill system Tue, 14 December 2010 23:28
I can see why relative 'skill' of players is a factor for calculating points for chess and perhaps (though I'm not so sure) for a more luck-based game like TTR, because all players start on a level playing field.

M44 However, is HEAVILY influenced by the relative balance of forces and card hand size. On top of THOSE you still have to add dice luck and card-draw. Therefore the relative skills of the players should be a relatively MINOR part of any calculation of merit points after the fact.

That is why I suggested that the points should be calculated almost exclusively on the difficulty of the scenario for each side.

The way things are calculated, it HEAVILY favours players of lower (known) skill, and could badly affect the way higher ranked players SELECT their opposition. But a new player of unproven skill could be a terrific and experienced player. Why should they be favoured by the calculation?

If the table is not to be based on MERIT, what is the point of its existence?

[Updated on: Tue, 14 December 2010 23:30]

      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Discussion about the Skill system Wed, 15 December 2010 04:48
Brummbar44 wrote on Tue, 14 December 2010 14:50

Sam makes some great points but I have to disagree here...to a certain degree.

I've been blown out on more than a few games because the dice simply fail. So a 6-1 loss with a hit percentage of 30% to the winners say 50% should earn them more points?! The game wasn't truly reflective of my abilities, just poor luck...should I additionally suffer because of that? It's frustrating enough to know that the dice are sour, knowing I'm going to lose more points as a result would be a turn off.

I would think that if the ranking system is to survive, the rolling percentage should be taken into account.

Malcolm, we've both played enough games to have plenty of tragic stories like that. I feel your pain. (In that Op Cobra game last night, I played an Assault Center to kill the two German armors ... and instead rolled 7 consecutive flags, retreating them to utter safety. Shocked )

It may be possible to estimate what percentage of a player's score in a game was due to luck vs skill, but there are so many factors to consider: Should you have rolled against the infantry instead of the armor? What were your dice averages against inf vs tanks vs arty, and eventually vs planes? How do you factor in cards or attackers where stars are hits? How do you factor in the odds for no-retreat situations? What about the bad luck of the cards? Of should you have played your cards differently, which might have prevented the bad luck?

I would tend to view it another way: Every once in a while, you'll have bad luck -- a hand where the cards are leave you no possible defense, or a game where the dice turn totally against you. Like a cost of doing business. If that happens in, say, 5% of games, just recognize that 5% of the time, you'll lose and it won't be your fault. And 5% of the time you'll win, because your opponent had inescapably bad luck. And if it happens to you (a strong player), it's probably happening to the other strong players about 5% of the time, too. If you play a decent number of games, by the law of large numbers, it'll average out. Sometimes you'll lose those, sometimes you'll win them.

In the log run, it would have a slight damping effect on ratings. High-rated players would tend to get pulled down a little occasionally, low-rated players would tend to get pulled up a little occasionally.
      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Discussion about the Skill system Wed, 15 December 2010 05:44
Yann wrote on Tue, 14 December 2010 09:59

Thanks sam1812 for your excellent comments. All your points make a lot of sense! Very Happy Maybe it is just a matter of tweaking the values and putting more inertia in the system?

Still, the non symmetrical property of the ELO system bothers me.

Obviously, I need to think about all this some more... Rolling Eyes

Yann

Thanks, Yann. I'm glad to be helpful. I do suspect that just some minor tweaks will make a big difference.

By "non-symmetrical," do you mean that if you play a two-game match against somebody, identical results in the two games won't return the two players' ratings to where they started? I had noticed that, too. Here's a suggestion for that. Insert a rule that if two people play the reverse side of a scenario against each other before either of them plays another game, then the points for both games will be calculated based on their ratings as they stood before game 1. (But if they then play the scenario a third time, start a fresh calculation -- don't keep looking back to game 1 or 2.) Easy to implement, and doesn't require a lot of record-keeping.
      
Axelb9
Junior Member
Major

User Pages
Posts: 25
Registered:
November 2010
Re:Discussion about the Skill system Wed, 15 December 2010 06:53
Or as an even simpler measure apply skill only after the two game full match has been finished based on cumulative medals. I do not see the reason why would we need to apply skill to lopsided single games at all. That's not the way the original game was designed. One should still be let play single games of course but they should count as ranked.
      
Pike Bishop
Junior Member
Arnhem Victory

User Pages
Posts: 16
Registered:
January 2005
Re:Discussion about the Skill system Wed, 15 December 2010 15:14
Which matches my thoughts Axelb9, except that the single games have the potential to count towards the skill rating if and when the player eventually plays the same scenario as the opposite side. At that point the results of both games are used in the calculation.

[Updated on: Wed, 15 December 2010 15:15]

      
Yann
-= Crew =-
KS Backer - Skeleton

User Pages
Posts: 2352
Registered:
October 2002
Re:Discussion about the Skill system Wed, 15 December 2010 18:33
sam1812 wrote on Wed, 15 December 2010 05:44


By "non-symmetrical," do you mean that if you play a two-game match against somebody, identical results in the two games won't return the two players' ratings to where they started? I had noticed that, too.

That's correct. It is due to the non-linear property of ELO. I was wondering if we could get rid of the "bell curve" constraint and still have something meaningful. Some food for my thoughts I guess.

Quote:

Here's a suggestion for that. Insert a rule that if two people play the reverse side of a scenario against each other before either of them plays another game, then the points for both games will be calculated based on their ratings as they stood before game 1. (But if they then play the scenario a third time, start a fresh calculation -- don't keep looking back to game 1 or 2.) Easy to implement, and doesn't require a lot of record-keeping.

Actually, what you are suggesting (like several other people did lately) is not easy to implement. The server manages hundreds of unrelated games in the same short time period. So it would require:

- at the end of a game, store the Skill values of the two players in a database table (you'll see why in a moment).
- search for an earlier game with the same scenario and players. Make sure that this game was not already part of another "pair" (which implies storing this information in a second database table).
o If you found a game, undo the Skill changes that were done then (using the first table), add up the two scores, apply the ELO formula and update the players Skill values. Since you found a "pair", mark the two games together by inserting a record in the second table.
o If you did not find a game, apply the ELO formula and update the players Skill values.

As you can see, it's complicated and does require bookkeeping and database storage space (something we are very cautious about).

Generally speaking, having to look back in the past to undo or redo things is a bad idea... Confused

So I'd rather come up with a linear formula Wink

Yann
      
Brummbar44
DoW Content Provider
Artillery Specialist

User Pages
Posts: 1129
Registered:
June 2004
Re:Discussion about the Skill system Wed, 15 December 2010 18:48
Here is a system I developed a long, long time ago for Vassal and just never did get around to making it happen. However, I believe the concept is still sound and you may be able to extract parts of it or re-formulate to make it work here (Karma could take the place of Prestige).

Anyway, thought it might be worth the post...

Cheers!


Calculating Rank:

There are three factors that are taken into consideration when calculating rank and some matter more than others. There are Experience, Skill and Prestige and are ordered in that importance.

Experience - This is the most important factor in determining your rank. It can only increase (you can't become less experienced although you still will make mistakes, believe me!)

You gain a +5 factor for experience for each game you log against another Brummbar registered opponent on the VASSAL system.



Skill - This factor can fluctuate quite a bit. It can go up and down depending on the skill level of your opponent and how well you did during a game. It can help get you promoted and just as easily get you demoted. Your Skill starts out at 1 and can never drop below that. This factor is a multiplier to your Experience so it makes up a large part of your Rank.

Skill Scoring System :

Over/Under - This is the basis of the system. Medals earned are compared to the avg. medals earned for that side on that scenario. The avg. medals earned is rounded down. If the medals you have earned are equal to or Over then you will get a score of +1. For each additonal medal over the average you add and additonal 1. If the medals you have earned are under the avg. then you receive a -1 score. For each medal you are under you receive a further -1 score. Further, if you are 'over' and your opponent is 'under' you may get more skill points as the two skill levels are now compared (if he was more skilled than you at the begining of the game). All players start out with a Skill score of 1 and may never drop below that no matter how poorly they play.

Formula (see calculation examples below):

Current Skill Level + Equal to or Over the Average for the Scenario + Skill Difference (if Opponent Under & higher Skill than you up to a mximum of +5) = New Skill level
or
Current Skill Score - Under for the Scenario = New Skill Level



Prestige - This is largely a reflection of your playing ettiqutie and can be determined not only by your opponent but also by some built in factors (ie. If you are a highly experienced player and quite skilled but take the Axis side in the Omaha scenario against a rookie, that is just plain wrong and will be calculated automatically based on this info). Presitge points can be earned by being courteous and timely when playing a VASSAL game (basically, don't screw your opponent around during the game and he will give you positive prestige points). These points may be argued in rare circumstances.

Prestige Scale;
+20 Rank Points for Exceptional courtesy and timely play.
+10 Rank Points for a Good Game
- 10 Rank Points for being difficult
- 30 Rank Points for Playing a lesser skilled player on a scenario where the average medals spread is two or more (even if the lesser skilled player requested the game, if that is the case and they twist your arm they can always award your +10 or +20 still) dropping a game, being very difficuly, unreasonably slow etc...
- ??? or Discharge if player has multiple violations (subject to review by court martial).



Each time you play a VASSAL game these points are tallied and may enable you to get promoted (or demoted if you lost skill as a result of the game). What does promotion get you? Well...it might do a few things. First, you get to look cool as you get a rank badge beside your name and people will respect your level of play (or hunt you down...more liklely) but more importantly it will help determine who plays in a tournament or, if it ever happens, help establish rank for an Overlord game.

The formula for calcluating Rank is; Experience x Skill + Prestige = Rank



The Ranks:

Much like in the real army, the VASSAL army has two rank systems - The Men and the Officers. Everyone starts out as a Private and will earn their way up the ranks. At some point you will earn your commission and become an officer. Once this happens you can never be demoted to the 'Men' ranks regardless of your skill score. In fact, on achieving Officer status your skill score and experience is reset to 0 (?).

Men

Private : 1 - 300 Rank Points or RP

Corporal : 301 - 600 RP

Sergeant : 601 - 900 RP

Sergeant Major : 901 - 1200 RP

Chief Warrant : 1201 - 1500 RP



Officers

Lieutenant : (1501) 1 - 500 RP

Captain : 501- 1000 RP

Major : 1001 - 2000 RP

Colonel : 2001 - 5000 RP

General : 5000 - 10 000 RP

Field Marshall - 10 000 + RP






Calculating Rank Example:

Brummbar (allies) V Lisa G (axis) - Pegaus Bridge
Allies: 4 won - (3.6 avg. round down to 3) = +1 for match, +1 for one medal over
Axis: 2 won - (2.5 avg. round down to 2) = +1 for match

New Skill levels; Brummbar = 1 + 2 = 3, Lisa G = 1 + 1 = 2 (Every player starts with a Skill of 1)

Rank Points for Brummbar is 5 (for the game) x 3 (for Skill) + 10 (for prestige) = 25 rank points.

Brummbar (axis) V Lisa G (allies) - Pegasus Bridge
Allies; 4 won : 3 avg. = +2
Axis; 1 won : 2 avg. = -1 <-- Because Axis is 'under' Allies player gets to compare Skill scores
Lisa G = 1 : Brummbar = 2 ...there is a difference of 1 so the Allies player adds that to their total for this game...
+2 for medals +1 for skill diff = +3 to Skill for Lisa G
-1 medals for Brummbar

New Skill levels; Brummbar = 2, Lisa G = 5

Rank Points for Brummbar is (5 (current experience) + 5 (for this game)) x 2 (for Skill) +10 (for prestige) = 30 rank points



Brummbar (allies) V Lisa G (axis) - St. Mere Eglise
Allies; 3 won : 3 avg. = +1
Axis; 4 won : 2 avg. = +3 -- +1 for matching, +1 for each after that is over.
Niether player is 'under' so skill scores aren't compared

New skill levels; Brummbar = 3, Lisa G = 8

Rank Points for Brummbar is (10 (current experience) + 5 (for this game)) x 3 (for Skill) + 10 (for prestige) = 55 rank points

Brummbar (axis) V Lisa G (allies) - St. Mere Eglise
Allies; 1 won : 3 avg. = -2 <-- -1 for being under the avg. and -1 for beign an additional medal under
Axis; 4 won : 2 avg. = +3
Lisa G = 8 : Brummbar = 3 ... there is a difference of 5 which is added to Brummbar's total
+3 for medals, +5 for skill diff = +8 to Skill for Brummbar
-2 Skill for Lisa G

New Skill levels; Brummbar = 11, Lisa G = 6
Rank Points for Brummbar is (15 + 5) x 11 +10 = 230 rank points



Brummbar (axis) V Lisa G (allies) - Omaha Beach
Allies; 1 won : 3 avg. = -2
Axis; 6 won : 5 avg. = +2
Lisa G = 6 : Brummbar = 11 ... The difference is negative, the higher skilled player is expected to win so no bonus here...it's penalty enough that the lower skilled player is going to lose 2 skill points.
+2 Skill for Brummbar
-2 Skill for Lisa G

New Skill levels; Brummbar = 13, Lisa G = 4
Rank Points for Brummmbar is (20 + 5) x 13 -30 (Brummbar shouldn't have played the Axis against a lesser skilled player on Omaha beach, any scenario where one side has a two or more average spread is condsidered very bad ettiqette) = 295 rank points



Brummbar (allies) V Lisa G (axis) - Omaha Beach
Allies; 1 won :3 avg. = -2
Axis; 6 won :5 avg. = +2
Lisa G = 4 : Brummbar = 13 = +9 to Skill for Lisa ...but maximum points for skill difference can only be +5
+2 for medals, +5 for skill diff = +7 to Skill for Lisa G
-2 Skill for Brummbar

New Skill levels; Brummbar = 11, Lisa G =11
Rank Points for Brummbar is (25 + 5) x 11 +10 = 340 rank points Brummbar is now Promoted to the rank of Corporal!



Brummbar (axis) V Johnny (allies) St. Mere Eglise
Allies; 4 won : 3 avg. = +2
Axis; 3 won : 2 avg. = +2

New Skill levels; Brummmbar = 13, Johnny = Base of 1 + 2 = 3
Rank Points for Brummbar is (30 + 5) x 13 +20 (Brummbar was very courteous and played the right side giving a lesser skilled player a chance) = 540 rank points

Rank Points for Johnny is 5 x 3 +10 = 25 rank points



Brummbar (allies) V Lisa G (axis) Operation Spring
Allies; 3 won : 3 avg. = +1
Axis; 6 won : 5 avg. = +2

New Skill levels; Brummbar = 14, Lisa G = 13
Rank Points for Brummbar is (35 + 5) x 14 +10 = 570 rank points



Brummbar (axis) V Johnny (allies) Pegasus Bridge
Allies; 4 won :3 avg. = +2
Axis; 0 won : 2 avg. = -2 <-- Johnny kept a skilled player 'under' and will score big skill here.
Johnny = 3, Brummbar = 13; Skill difference of +10 goes to Johnny...but the maximum from Skill diff is 5 so adjust to +5
+2 for medals, +5 for Skill = +7 to Skill for Johnny
-2 to Skill for Brummmbar

New Skill levels; Brummbar = 12, Johnny = 8
Rank Points for Brummbar is (40 + 5) x 12 +10 = 550 rank points

Rank Points for Johnny is (5 + 5) x 8 +10 = 90 rank points.
      
TheBigEik
Junior Member
Major

User Pages
Posts: 6
Registered:
November 2010
Re:Discussion about the Skill system Thu, 16 December 2010 00:41
Well. I dunno if this is suggested b4, I have no time to go through the entire thread, but I strongly suggest that every overperformance is rewarded with points.. so if you get 4 flags on a game that averages 3 flags for the side you are playing you should always get points.. even if you lose! Reason is that you overperformed!! Simple as that..


This will stimulate that experienced players will introduce noobs to the game without having to think too much about losing points!
      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Discussion about the Skill system Thu, 16 December 2010 03:13
Yann wrote on Wed, 15 December 2010 12:33

sam1812 wrote on Wed, 15 December 2010 05:44

Here's a suggestion for that. Insert a rule that if two people play the reverse side of a scenario against each other before either of them plays another game, then the points for both games will be calculated based on their ratings as they stood before game 1.

Actually, what you are suggesting (like several other people did lately) is not easy to implement. The server manages hundreds of unrelated games in the same short time period. So it would require:

((Details omitted for brevity.))

As you can see, it's complicated and does require bookkeeping and database storage space (something we are very cautious about).

I think my suggestion may be simpler than it appeared at first glance. I specifically said, "before either of them plays another game."

You already have a Player table in the database. Just add a few simple fields to that table: 'Last scenario played', 'last opponent', 'side played in last game', 'rating before last game', and maybe one or two others that I'm not thinking of at the moment. When two people play a game, check to see whether each of them was the other one's last opponent on the opposite side of the same scenario. If yes, then use their prior ratings to calculate game 2's rating impact as a symmetrical 2-game match. If either person's previous game doesn't match, then simply don't count it as a "match," and enter the data on this game as the new last game played for both of them.

Otherwise, if the system had to track an unlimited number of games at an unlimited number of scenarios, for an unlimited number of players, I agree, the overhead in data storage and calculation could get really messy.

Using this approach, flipping the scenario immediately, or the next time you're online, will award the points symmetrically. And if one of the people played some games in between, so that the rating system gives slightly asymmetric results, it'll still tend to be pretty close. Smile
      
GoboGobo
Member
Major

User Pages
Posts: 48
Registered:
April 2006
Re:Discussion about the Skill system Thu, 16 December 2010 12:12
For a more detailed analysis of how the ranking performs, I'd like to be able to see a more detailed history of how my ranking changed over the last few games. I can see how I score, but not how my ranking changed.
Also, it would be nice if that was somehow tied in with the memoir44-online.com website.


Thinking about the skill system more theoretically, I think the basis is wrong as it is currently. You should not be looking at whether you won or lost the scenario, but you should be looking at whether you beat or lost from the average score for that scenario. That has to be in some way tied to the difference in ranking between the two players as well of course.

I mean, if I lose a scenario 5-4, from a worse player while on average it's 4.5 - 2 then it seems unfair that I lose ranking.
      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2257
Registered:
August 2006
Re:Discussion about the Skill system Thu, 16 December 2010 14:03
A number of people have raised the issue that it doesn't seem fair to lose points if they perform better than average. I absolutely agree that there should be some benefit for "overperforming." But that benefit doesn't have to be winning extra rating points. (If you were playing poker and you had a flush, but someone else had a full house, would you expect a share of the pot because your hand was better than average?)

By using zero-sum payoffs, the weighted average of all player ratings will always be 1500. A year or two from now, it won't rise to 1700 or 2000 or 2500.

I think people will feel a lot better about losing points for a loss if the payoffs are adjusted more heavily based on the number of medals. Using numbers from Yann's recent PDF (for a game between two 1500 players), if the Allied player wins Omaha 6-5, he currently gains 5 points -- but if one die rolls differently and he loses 5-6, instead, he currently loses 27 points. If the payoffs were more aligned with the chances, he should only lose 2 or 3 points, instead.

So when playing a match at Omaha, a strong player would hope to roughly break even rating-wise as Axis, and earn some points by overperforming as Allies.
      
Brummbar44
DoW Content Provider
Artillery Specialist

User Pages
Posts: 1129
Registered:
June 2004
Re:Discussion about the Skill system Thu, 16 December 2010 16:26
sam1812 wrote on Tue, 14 December 2010 19:48



I would tend to view it another way: Every once in a while, you'll have bad luck -- a hand where the cards are leave you no possible defense, or a game where the dice turn totally against you. Like a cost of doing business. If that happens in, say, 5% of games, just recognize that 5% of the time, you'll lose and it won't be your fault. And 5% of the time you'll win, because your opponent had inescapably bad luck. And if it happens to you (a strong player), it's probably happening to the other strong players about 5% of the time, too. If you play a decent number of games, by the law of large numbers, it'll average out. Sometimes you'll lose those, sometimes you'll win them.
.


This sounds correct in theory. However, I for one am certainly beating your 5% odds here. I consistently am rolling in the lower to mid 30 percentile every game...while my opponents (not taking anything from them, they play well) are getting the high 40 percentiles every game!

How is that a measure of 'skill'?

Sorry for venting, but I'm getting tired of losing to the dice.
      
GoboGobo
Member
Major

User Pages
Posts: 48
Registered:
April 2006
Re:Discussion about the Skill system Thu, 16 December 2010 16:37
sam1812 wrote on Thu, 16 December 2010 14:03

A number of people have raised the issue that it doesn't seem fair to lose points if they perform better than average. I absolutely agree that there should be some benefit for "overperforming." But that benefit doesn't have to be winning extra rating points. (If you were playing poker and you had a flush, but someone else had a full house, would you expect a share of the pot because your hand was better than average?)

By using zero-sum payoffs, the weighted average of all player ratings will always be 1500. A year or two from now, it won't rise to 1700 or 2000 or 2500.

I think people will feel a lot better about losing points for a loss if the payoffs are adjusted more heavily based on the number of medals. Using numbers from Yann's recent PDF (for a game between two 1500 players), if the Allied player wins Omaha 6-5, he currently gains 5 points -- but if one die rolls differently and he loses 5-6, instead, he currently loses 27 points. If the payoffs were more aligned with the chances, he should only lose 2 or 3 points, instead.

So when playing a match at Omaha, a strong player would hope to roughly break even rating-wise as Axis, and earn some points by overperforming as Allies.


I would be careful with the zero-sum argument. ELO is based on zero-sum, yet on chess sites and chess ratings you will see that average rating will be above 1500.
Stronger players tend to stick around, while weaker players have more of a tendency of leaving.

ELO is not that good in coping with that, ratings tend to get a little bit inflated over time, and players then tend not to want to play with lower-ranked players, making it harder for newcomers to rise in the rankings.

[edit]
Actually, what I'm saying here about ELO is not true. ELO is not zero-sum. but still, having a zero-sum rating system where players come and go does not help you one bit.

[Updated on: Thu, 16 December 2010 23:11]

      
Pages (3): [1  2  3  >  » ]     
Previous Topic:Armor Excellence
Next Topic:the first Major General
Goto Forum: