Probability of being dealt ten tiles of one suit?

Japanese Reach Mahjong Rules. Strategy, news, sets - anything!

Moderator: Shirluban

Post Reply
Goldeneye
Senior Reacher
Senior Reacher
Posts: 135
Joined: Tue May 17, 2011 6:35 pm

Probability of being dealt ten tiles of one suit?

Post by Goldeneye » Mon Jul 14, 2014 2:11 pm

Just now on JRM, I was a ko player and got dealt ten tiles of one suit - I am guessing that the probability of that isn't very high...

Iapetus
Senior Reacher
Senior Reacher
Posts: 195
Joined: Sat Jul 16, 2011 1:00 am

Re: Probability of being dealt ten tiles of one suit?

Post by Iapetus » Mon Jul 14, 2014 3:29 pm

I think the probability would be

(3*(36 choose 10)*(100 choose 3))/(136 choose 13)

=0.00025488...

Or about 1 in 4000.

Senechal
Senior Reacher
Senior Reacher
Posts: 175
Joined: Wed Mar 04, 2009 5:42 am
Location: The frozen part of HELL!

Re: Probability of being dealt ten tiles of one suit?

Post by Senechal » Mon Jul 14, 2014 5:37 pm

Iapetus wrote:I think the probability would be

(3*(36 choose 10)*(100 choose 3))/(136 choose 13)

=0.00025488...

Or about 1 in 4000.
Iapetus' number for chin'itsu-grade hands sounds good. Conversely, a hon'itsu hand of 10/13 tiles would be:

(3*(64 choose 10)*(72 choose 3))/(136 choose 13) == 5.6 % (one in 19)

However, this is an approximation as there are cases that overlap, with some of the hands that have more than 7 jihai (alone, that is 0.468 %).

Referee
Expert Reacher
Expert Reacher
Posts: 382
Joined: Sat Apr 24, 2010 6:24 pm

Re: Probability of being dealt ten tiles of one suit?

Post by Referee » Mon Jul 14, 2014 6:52 pm

What's the 3* for?

Senechal
Senior Reacher
Senior Reacher
Posts: 175
Joined: Wed Mar 04, 2009 5:42 am
Location: The frozen part of HELL!

Re: Probability of being dealt ten tiles of one suit?

Post by Senechal » Mon Jul 14, 2014 9:56 pm

You have to set your exclusions.

One suit vs rest.
Second suit vs rest.
Third suit vs rest.

In my example, it then becomes :

One suit and honours vs rest.
Second suit and honours vs rest.
Third suit and honours vs rest.

In both cases you have to add those together, thus 3x. In the hon'itsu case, there is some overlap.

User avatar
or2az
Gold Boarder
Gold Boarder
Posts: 823
Joined: Tue Aug 20, 2013 5:10 pm
Location: USA

Re: Probability of being dealt ten tiles of one suit?

Post by or2az » Mon Jul 14, 2014 11:22 pm

I would like to follow these mathematical calculations but I am not familiar with the usage of the word "choose" in this context.
Can you be more specific with the actual numbers you are multiplying and dividing in these calculations.

User avatar
Shirluban
Moderator
Moderator
Posts: 780
Joined: Thu Jan 22, 2009 1:53 pm
Location: Svartalfheim
Contact:

Re: Probability of being dealt ten tiles of one suit?

Post by Shirluban » Tue Jul 15, 2014 12:14 am

By "choose", he means Combination.

"64 choose 10" = "number of possible sub-sets of 10 distinct elements from a set of 64 elements" = 64! / (10! × (64 - 10)!)
Cats don't do タンヤオ (tan-yao) but タニャーオ (ta-nya-o).
World Riichi Championship Rules 2022
Comparison of riichi rules around the world

Senechal
Senior Reacher
Senior Reacher
Posts: 175
Joined: Wed Mar 04, 2009 5:42 am
Location: The frozen part of HELL!

Re: Probability of being dealt ten tiles of one suit?

Post by Senechal » Tue Jul 15, 2014 12:15 am

Choose refeers to stuff in combinatorics where out of Y tiles, you want X amount, "Y choose X" (usually written with two-line parthenses, Y on top of X) is a simplified mathematical representation of the factorial operations Y! / (X! * (Y-X)!). Each factorial number is the product of multiplying every integer from 1 to itself ( 1 * 2 * 3 * 4 * ... * N ). 6 choose 4 == 6! / (4! * 2!) == 720 / (24 * 2) == 15.

A single choose operation is symmetrical, as choosing 4 of 6 is the same as "not choosing", or "choosing to negate" 2 of 6. It's easy to visualize choosing 2 of 6 items as being 6 * 5 {*4*3*2*1} / 2 {*4*3*2*1} == 15, but the same formula as above applies. When combined with other choose operations or operations, then things get funky.

The other issue with factorials as that they explode in value faster than exponentially. The math curves would be ranked x^x > n^x > n! > x^n. A million cubed is 10^18. A million factorial is 5.56 million digits long. Still, with only 136 tiles to worry about, that leaves numbers to be calculated with systems able to accept values in the 136! range... of 3.659 * 10^232, without rounding them arbitrarily to values.


To be fair and answer your question, writing out every value that the choosing gives is both exhaustive and trivial, because a lot of it divides itself out of the way.

User avatar
or2az
Gold Boarder
Gold Boarder
Posts: 823
Joined: Tue Aug 20, 2013 5:10 pm
Location: USA

Re: Probability of being dealt ten tiles of one suit?

Post by or2az » Tue Jul 15, 2014 12:45 am

Senechal wrote: "Y choose X" is a simplified mathematical representation of the factorial operations Y! / (X! * (Y-X)!).
This definition says it all. I am very familiar with the term factorial. Thanks.

I assume that this calculation still takes a lot of button pushing on a calculator even though some of it divides itself out of the way......or is there now a really quick shortcut available that I don't know about.
3*(36 choose 10)*(100 choose 3))/(136 choose 13)=0.00025488...

(3)(36!)(100!)(13!)(123!)
-------------------------------
(10!)(26!)(3!)(97!)(136!)

Amazing how a seemingly simple question concerning a probability turns into an educational math lesson. Great Stuff!
(Anyone want to learn how to solve 1 equation with 2 unknowns.......just kidding....er....maybe)

Kyuu
Expert Reacher
Expert Reacher
Posts: 273
Joined: Fri Sep 23, 2011 2:27 pm
Location: Chicago, IL
Contact:

Re: Probability of being dealt ten tiles of one suit?

Post by Kyuu » Tue Jul 15, 2014 8:22 pm

or2az wrote:Amazing how a seemingly simple question concerning a probability turns into an educational math lesson. Great Stuff!
(Anyone want to learn how to solve 1 equation with 2 unknowns.......just kidding....er....maybe)
Mahjong probability is anything but simple because the probabilities themselves are dynamic.

Post Reply