Archived from groups: rec.games.mahjong (
More info?)
Martin and Cofa,
Firstly, I cite some paragraphs of your thoughts from this thread:
> >> I doubt if a handicap system is any good. Handicaps are useful in
> >> games where skill of of the greatest importance. Mahjong has its own
> >> handicap: the way the winds will spread the tiles amongst the players.
> >> Even newcomers can beat experienced players.
> However, in mahjong there are no absolute terms to tell a good player
> from a bad one. Everyone knows a beginner can beat even the most
> experienced mahjong master. Because of the luck that is involved
> (wherefor I used the metaphore of the winds). We have concluded before
> that mahjong is a game of skill, luck [and intuition]. But, eventually,
> fate determines whether a players gets a handicap (bad tiles) or not
> (good tiles).
Secondly, I cite a part of original message (1st in a thread):
> I received an inquiry from Marvin:
> "Does MJ has a handicap system? Could one be developed much like the golf
> handicap system?"
Now, let's look at definitions of a word "handicap" as I have found in
the web:
http://www.thefreedictionary.com/handicap
handicap - advantage given to a competitor to equalize chances of
winning
http://golf.about.com/cs/golfterms/g/bldef_handindex.htm
Definition: A USGA Handicap Index is a numeral, to one decimal place,
that represents a golfer's potential for scoring. A handicap index of
14.5, for example, indicates that a golfer will, on his or her best
days, shoot somewhere around 14 or 15 strokes over par.
As I personally see we have several issues in here:
* how to differentiate/ equate different skills level of players in a
game?
* how to measure/ quantify those inequity of chances to win? (!
Classification and Ranking ?)
* given some quantitative measure of luck (like golf handicap index)
what are PROBABILITIES of a player to win, to get, say, second place
in 16-hands match, etc.?
Golf index clearly says to me: if one player has index, say, by 3
higher that means that this player would finish event with
ON-SOME-AVERAGE 3 strokes more.
If we look at Ratings in chess (Elo coefficients) that system works
(is designed to work in such a way!) like:
when difference of Ratings of two players is 200 points that means
that, say, in 10-board match their score would be AROUND 7.5 to 2.5
that is 3:1. (As clear consequence 400 points means 9:1 ratio, 100
points means 1.7320508 (square root of 3):1 ratio, etc.).
Game of Go. 100 points means that players would be "equal" with one
stone of HANDICAP. Look, that game is maybe the only perfect example
of "internal" handicapping.
We can easily convert that way of thinking to 16-hands matches in MJ:
based on pre-defined or pre-calculated Scores/ Indices of 4 players
one can calculate on-the-average final raw scores and hence places.
That clearly leads to Classifications and Ranking. I tried to start
that discussion on EMC-2005 forum:
http://www.mahjongnews.com/phorum/read.php?f=1&i=129&t=129
Being statistician and having experience on tournament holding I CAN
prepare (on a volonteer basis) some kind of reference tables: Ranking
==> Index ==> Probabilities ==> Scores ==> Places ==> Ranking. Or, at
least, I know how to do that. Major stuff concerning grounds to do
that is already mentioned in above mentioned forum.
As for luck (and intuition) in general I would give an example of
tennis.
Let's assume that two player's level is so that chances to win a ball
are 9:1 when better player serves, and 7:3 when the other player
serves. Those numbers only say that out BIG number of serves winning
would break in proportions close to the above mentioned. Though less
skilful player can win even several balls (no matter who serves) in a
row.
The way to decrease that statistical "uncertainty" of "who is the
king?" is to play longer events (or to assigned any winning scores
ONLY after big number of "tries"). Like in tennis we have system
Ball-Game-Set-Match. On the longer event there are more CHANCES to
reveal winner (or even try to quantify results-probabilities etc.).
In MahJong we have similar thing: Hand-(Round?)-Match(16 hands). From
my personal view any Rating and Ranking should be applied to those
16-hands matches. We can try to define typical ranges of Raw Scores
(Points) and Victory Points (or VP, those points for places based on
4-2-1-0 system).
As a possible summary: anything goes in ONE particular MJ hand. It may
depend on Honors distribution, Rules, condition of players, time of a
day, etc.
Though: in a long event Master have more (all?) chances to reveal
his/her level of play.
Hope to continue on Statistical project,
Vitaly Novikov