Players Skill Level

Many players want to know other players skill levels while they are playing against others or even before the game. That is reasonable as this knowledge will significantly enhance their chance of winning. For those who wants to know more about classification of player's skill levels, please see my another posting - 'Risk Strategy - Knowing yourself and knowing your opponents'. However, it does not provide you much information on out of game research on other players. You may use information such as 'Player Ranking' and 'Skill Level' that are publicly available to determine a player's skill level. They are indeed not a good skill indicators. Here are why: 1) Player Rank: it is not a good indicator because the way a player obtains his/her tokens. Every player gains 20 token per game regardless win or lose. Therefore a player who has played 1000 games with 0% winning rate would have 20000 tokens and his/her rank would be a [u]General[/u], the highest rank indeed. How skillful he is comparing to a player who has only played 100 games with 50% winning rate. I'm not able to tell, however, one thing I am certain the former player played more games on this site than the later. Therefore it really should be called: Player's Experience (this site only). 2) Skill level: This sounds to be the indicator of a player's skill level. I think Vexer would be able to fill us in how this is calculated and why this is a good indicator of player's skill level. Personally I do not believe it is accurately reflecting players' skill levels. In my opinion, to accurately show a player's skill level, there should be 3 distinct indicators: [list type=decimal] [*]Experience [*]Historical (Overall) Skill Level [*]Recent Skill Level [/list] [list type=decimal] [*]Experience - this can be the existing 'Player Rank' [*]Historical Skill Level -a formula must take into account of the following factors to accurately reflect the true skill levels of players: [list] [*]# of Players in a game: this avoids the advantage of a player who only involves 2 player games over a player who plays mostly 8 player games [*]Opponent's Skill level: this avoids the advantage of a player who mostly played with low skill level players [*]Missed Turn Games: Whether missed turn games should be counted against a player's skill level is debatable. My personal opinion is they should not. A more controversial, however, is whether the winner should be getting the full credit for winning a game with skipped players. It is hard to determine the impact accurately as there are 2 factors may distort the credibility of the winner, they are: a) Whether a player misses his/her turn at the early stage of a game or later stage b) # of Players missed their turns in a game The problem is if you cannot determine the impact, you cannot determine the credit the winner should get. [/list] [*]Recent Skill Level - it has the same formula as the historical skill level, except this only take into account of most recent games. The size should be statistically significant, say, at least 30. I would suggest 100 most recent games if possible. [/list] The reason why I think there should be three distinct indicators, because each indicator shows a unique aspects of a player's skills and they are not convertible from one to another, e.g. 1000 games experience = 10 skill points? If you don't know the conversion ratio, you cannot combined them into 1. If all 3 indicators were available, when I do a research on a player, the first thing I look is the 'Recent Skill Level', which is the most up-to-date skill indicator, then I will look at the experience level, because it would tell me if this player is likely to make level 0 and/or level 1 players' mistakes during the game. The historical skill level is a good indicator to compare 2 players with similar recent skill levels and experiences.

Removed

Removed

I think the official dominating12 list is a nice way to see hwo good ppl are. http://www.dominating12.com/forums/dominating12june12.html Of course, measures get more accurate when a player plays more games. For example, if you are a newby, and in your first game, against Vexer, Fendi and Holt, you win. Well, all good formulas will tell you that you are a great player. However, it could be both skill, or pure chance. So for players who played only little games there is no real good option but to watch how they play. Also, the way I check new players is simply check their tallies. Who did they play against, and who did they beat. So if a newbie played 5 games against Fendi, and won them all, its probably a good player. However, if he only beat ppl I never heard of, it might be quite a newbie, even if he has a good rank :)

If I'm not mistaken, the skill points has combined experience points which I would argue if it should be combined. I have stated my reasons above why I oppose to it. Also if you don't have a recent skill level indicator, it is hard to accurately tell a player's recent skill level. A more experienced player will tend to have higher skill level which is not always true. By the way, if you want to know a player's skill level accurately, you need a larger sample size than 5 games. In a standard deviation distribution, you need at least 30 random games.

One more thing, why the Rank has to be manually purchased by players. If it is an indicator of experience, Rank should be automatically awarded. I can understand maps and game types, but really I have no incentive to rank myself up!

You can't become general with 0% winning rate. Yes, you need tokens to get higher ranks, but you also need points, which you only get by winning games. I, however, do agree that I am not that much of a fan of purchasing ranks with tokens, though as it is barely the only thing tokens can be spend on, I don't think it is that much of a problem and it also requires you to have played a minimum amount of games, so newbies cannot become a high rank after some "lucky wins".

Yeah. For example, I have 8000 tokens right now, but I don't have enough points for the next ranks, so I can't get it.

Thanks for pointing out that to me. I wasn't aware the ranking requires skill points. Nevertheless, the rank is still not a good indicator of skill level as experience and skill level should not be combined. Once it is combined, it neither indicates the experience nor the skill of a player. 2 extreme examples would be a) a player played 1000 games with 0 winning rate b) a player played 50 game with 100% winning rate, both players will be at very low ranking, but one with extensive experience and another with veteran skills. Also I don't believe there is universally accepted conversion rate which allows one skill to be converted to another and therefore they should be 2 different indicators. With the current setup of the Ranking, it takes the lower skill (experience vs Skill) as the ranking and therefore, it may be a good indicator for a high ranked players, but not so for low ranked players.