Ura Dora Strategy

[zzo38]

I wonder whether you could help with a technical issue. I\'ve been working on translating the Toupaiou software into English by editing the text straight up in Notepad++. However, it will not run because the filesize gets changed when the string/length is altered. Just switching the string for another of the same size, say, two kanji for two kanji and it will still run. Do you know a way around this problem? Can it be circumvented?

You can try using Resource Editor. It may or may not work, depending on the way the resources are stored in the file.

[HotelFSR]

zzo38:

Would have used resource editor, but the files are not structured in such a way that you can get at all the text.

This is why I had to resort to Notepad++.

Any ideas?



Garthe:

Assuming that more doras purely adds more variance (which i\'m not sure is actually the case, as discussed in the other post i started on red fives strategy) then it does not make sense that pros would want to play that way in order to make more money. Logic dictates that their greater earnings would be offset by greater losses (even when adjusted for whatever edge they may have). We can assume there is something else at play.

Now, if we accept that there is actually a skill element inherent in increased dora games (the randomness of ura-dora largely notwithstanding) and we consider that- in some ways- they may actually reduce variance (as discussed in the red fives post) then what you say makes a lot of sense.

Am i on the right track here?

[iandstanley]

Removing ura-dora from the game, in my opinion, wouldn\'t add to the strategy at all. The strategy in riichi mahjong is more focused on not dealing into opponents hands than making hands that are more valuable. After all, the tiles that come to you are random, and the hand you are making depends much on luck. On the other hand, not dealing into a riichi requires more skill. If ura-dora were removed, the threat of riichi\'s would diminish and so would the importance of defensive play.

I totally agree. If you want to reduce luck in an inherently luck based game. Key elements that must remain is a random draw. Keeping this if you are to reduce randomness then the next step after removing dora would have to be removing a suit. Removing the option of a kong would remove randomness as you would then only draw/claim one random tile instead of two .

Removing randomness is not always beneficial otherwise we would all still be engrossed in tic-tac-toe.

The whole basis of mahjong is to be able to quickly reformulate a strategy when randomness deals you favourable or unfavourable conditions. The skill is in:

• factoring in incomplete information from your opponents discards and recalculating odds of both your intended hand and the odds of the safest discards
• knowledge of the special hands
• As with poker, ... Knowledge of your other other players habits & behavioural quirks/body language

[HotelFSR]

It\'s not about removing luck, it\'s about creating a fine balance between luck and skill.

Several professional leagues in Japan, such as the JPML to which Garthe and Jenn belong, play league games with no ura-dora. They also don\'t use red fives or ippatsu. This is done to reduce excessive variance.

Removing ura-dora doesn\'t really add strategy so much as it reduces excessive luck. Personally I like playing both with and without ura-dora.

[Benjamin]

Garthe:

Assuming that more doras purely adds more variance (which i\'m not sure is actually the case, as discussed in the other post i started on red fives strategy) then it does not make sense that pros would want to play that way in order to make more money.

And you play poker for a living? I suppose you don\'t play much live games.

Having more variance in the game will cause the pros to win less over the long run, yes, but IN EACH GAME. If customers like playing with red fives and other inflate-rules, then the pro gets more action and wins more over the long run. I\'m sure a lot of pros would love to play customers at A-rules, but they wouldn\'t get much business if they tried.

You\'re right in that there\'s skill in adapting to red-fives and ura-dora. The fact that most people over here enjoy playing with those rules more is a lot more significant.

[HotelFSR]

I understand that. You may have misunderstood me.

It\'s no mystery that people prefer playing with all the extra \'luck\' rules, or that this has its own benefit for pros.

[Benjamin]

Assuming that more doras purely adds more variance (which i\'m not sure is actually the case, as discussed in the other post i started on red fives strategy) then it does not make sense that pros would want to play that way in order to make more money. Logic dictates that their greater earnings would be offset by greater losses (even when adjusted for whatever edge they may have). We can assume there is something else at play.

You may have misunderstood me.

What did I miss?

You said that if more doras lead to more variance (which is true) then "their greater earnings would be offset by greater losses."

I said that pros play games with increased variance, even though they have less of an advantage, because they get more customers to play with them and thus make more money. Their greater earnings are not offset by greater losses, its just that they don\'t win as many games. In fact, since they have more customers, they wind up winning more.

Now, if we accept that there is actually a skill element inherent in increased dora games (the randomness of ura-dora largely notwithstanding) and we consider that- in some ways- they may actually reduce variance (as discussed in the red fives post) then what you say makes a lot of sense.

Now this, on the other hand, I don\'t understand. Of course there is a skill element in increased dora games; otherwise the game would be pure luck. I don\'t think anyone is arguing this. As for how this leads to your next phrase--in some ways dora reduce variance--and how that relates to what Garthe said, well, the connection is unclear to me.

[iandstanley]

Assuming that more doras purely adds more variance (which i\'m not sure is actually the case, as discussed in the other post i started on red fives strategy) then it does not make sense that pros would want to play that way in order to make more money. Logic dictates that their greater earnings would be offset by greater losses (even when adjusted for whatever edge they may have). We can assume there is something else at play.

With all due respect to someone with more riichi experience. I think I disagree.

Traditional Mahjong (which I have played for most of 30 years) where everybody pays everybody there is less reliance on damage limitation/defense. I would typically play for the best hand given what I was dealt/drawn and my strategy would be to improve the hand continually even if I risked not going out. Often I would turn to defensive play only if my hand was going nowhere. I would get many hands that beat the winner who went out cheaply even when I was still a tile or two off going mahjong.

Whilst every experienced player/pro, keeps a careful note of how he/she could improve the hand, he/she is always aware when a hand is lost or as good as and plays for the draw when the situation dictates (late game, or when his/her hand suddenly loses value e.g. remaining tiles of a special self-drawn hand are lost)


With Modern Chinese, Ari Ari & Riichi rules require much more defensive play as even a Ron\'ed mid hand of 4000-8000 pts can kill off your chances of winning on the day.

With Red Fives/Ura-Dora you increase the penalty for the careless player. A Pro understands how not to get caught on a discard (or at least drastically reduce) by examining the discards of all players. It is this ability alone that brings his play up. Yes he/she can be excellent at analysing his hand for the best possible chance but it is his/her defensive play that minimises the effect of other peoples wins.

After all it is a zero sum numbers game. Knowledge of odds of getting a particular hand gives a slight improvement to your overall score or your chances/number of times you get a limit hand. But careful defensive play should reduce your losses in the vast majority of times.

If we took the current rules and raised the limit to 4x the prior limit. Obviously all limit hands would pay off more...but the pro would still come out on top. His percentage win rate would probably be around the same. But the potential payoff would be greater.

It is the same with Red 5s/ura dora. The pro would get a similar percentage of at least the same (or potentially much larger pot).

Because Red fives/Ura Dora increase the payout and may affect the strategy of the inexperienced player (e.g retaining a red five for too long) thus increasing the win rate of the pro slightly. Over sufficient games (not the regular game of 8+ hands which could have a skewed effect on the payoff ratio) but over the course of a month or more the effect of red fives/ura dora bonuses would at least even out in it\'s distribution so that the pro would get an even share of the bonuses. With an even share of the bonuses the effect on the full time pro would be minimal.

Inexperienced players will get drawn to playing the pros with the potential of a bigger payoff during a single game. An individual inexperienced player may come away with a big win but if you group all the inexperienced players who play that pro over an extended period the average pay off rate would remain the same in favour of the pro.

[HotelFSR]

I suppose I wasn\'t being very clear. I think we\'re all on the same page here.


Naturally pros get more business playing with the extra doras, and people like the possibility of larger payoffs or better chances at beating the pros on a given night.

I was simply suggesting that doras add both some luck and some skill to the game, rather than luck only. They can also reduce the variance in distribution of starting hand values because of the way the limit system works.

Now, if we accept that there is actually a skill element inherent in increased dora games

I didn\'t mean that doras are pure luck (variance), just that they add variance. Obviously nobody would argue that they turn the game into one of pure luck!

Never having played in a jansou I also didn\'t realize at the time of writing that post that doras could increase the final payout size. I had thought that only the uma money changed hands (which is a lot less interesting).

As for the extent to which more dora = more business and exactly how much more payoff this gives pros (e.g. three versus four red fives, with or without gold dora) I can\'t comment, never having been to jansou or looked at the numbers.

[Benjamin]

@iandstanley

I\'m not quite sure how to put this, but the Japanese Mahjong I\'m familiar with doesn\'t work the way in which you describe it. For the most part, what makes pros better isn\'t reading discards.

Adding red fives and ura dora decreases, not increases, the need for defense. This is because many more hands are worth winning, whereas without the bonus doras it\'s not worth it to push those same hands. Non-inflated \"A-rules\" mahjong played at a high level is incredibly defensive. And kind of boring.

@HotelFSR

Ok, I think I see your point more clearly now.

Being able to adapt to the extra doras is certainly a skill, but I\'m pretty sure that on the whole, there is more variance. We might have to let it go at that.

As for reducing the variance in distribution of starting hands, I\'m not sure how you figure this. Doesn\'t having more possible starting hands at a greater range of values mean the opposite?

\"Doras increasing the final payout size\" Where did this come from? Are you talking about the cash bonuses that some parlors award when you win a hand with a bonus dora/ippatsu in it? I think this is pretty negligible.

more dora = more business---I doubt there\'s a relationship here. Despite all of the possible variations of dora you can have, using 1 red dora of each suit and having ura and ippatsu is by far the most common setting. I don\'t know of a single parlor that allows you to play furii against a pro and *doesnt* use ura dora or red fives.

[HotelFSR]

Oh, I think the extra dora probably do increase variance overall.

As for reducing the variance in distribution of starting hands, I\'m not sure how you figure this. Doesn\'t having more possible starting hands at a greater range of values mean the opposite?

Not when you consider the point values and the limit system, i think. Hands tend to be worth more, but also more clumped around the 3-4 han region and 1 han is much more rare. It was talked about more in the red fives thread. And as you pointed out yourself, more hands are worth winning.




Are A-Rules boring at high levels? I haven\'t seen it played so I have no idea. I would have thought it could be pretty boring when played with lower level players, but that it would make high level games more interesting.

[iandstanley]

I suppose I wasn\'t being very clear. I think we\'re all on the same page here.

I also think we are arguing the same cause :laugh:

I was simply suggesting that doras add both some luck and some skill to the game, rather than luck only.

Agreed. It takes a more skillful player to play carefully and reduce the potential of getting clobbered.

They can also reduce the variance in distribution of starting hand values because of the way the limit system works.

Sure if we plotted two histograms of yaku and yaku with red fives/dora then it will show a marked shift to the right over a number of games.

Never having played in a jansou I also didn\'t realize at the time of writing that post that doras could increase the final payout size. I had thought that only the uma money changed hands (which is a lot less interesting).

Couldn\'t comment there .. I am assuming that the degree of final payout is based upon the degree of the thrashing by the winner. e.g. with the pro tournaments and most of the online sites seem to have a tournament/league points value that is calculated from the respective totals at the end of the game. You each start with a five figure total of points (typically in the 25k-27k range) and the game ends with one or more persons dropping below 0 .... yet the points awarded are in double figures according to the relative positions at the end of the game.

As for the extent to which more dora = more business and exactly how much more payoff this gives pros (e.g. three versus four red fives, with or without gold dora) I can\'t comment, never having been to jansou or looked at the numbers.

I have never been to a jansou. But splitting the gaming audience into three categories:

1. Non bettors ... would not play in a jansou except under carefully controlled environments (ie all friends and no money changes hands other than the table fee)

2. Non-expert bettors. As with any kind of gambling As skill increases the bettors gets more careful and considers the odds. The further away you move from the expert gambler the more you will increase the probability of the gambler going for either the easiest win or the long shot.

i. the easiest win is unlikely to hurt the pro during the course of the game. It is also counter-balanced with the fact that the beginner makes a lot more mistakes and/or shows too much info in his discards. Going for the easy win all the time will bring you a net loss as a beginner.

ii. the long shot. Again going for the big payout vs a pro is unlikely to get you any where because of your discards / mistakes. But there is still a chance... like betting on the 20/1 shot at the horses.

With this category of bettors. Red Fives / Doras increase the potential payoff reducing the requirement for the beginner to rely on the bigger limit hands and therefore the player will be attracted by this due to it\'s reduced skill requirement!

3. Expert bettors/pros ... I suspect the effect would be negligble due to the skill level of both the near-pro and pros playing

[iandstanley]

Assuming that more doras purely adds more variance (which i\'m not sure is actually the case, as discussed in the other post i started on red fives strategy) ...

I think over the course of a reasonable number of games the number of doras and red fives will probably follow the normal distribution in stats. If so it would increase the mean (bigger payoffs per hand) and reduce the variance slightly because you are still hitting a hard limit on the max number of doubles The change is distribution would be interesting with less smaller hands like the 1Ks but with the bulk of the doras being in the 1-3 range I would expect that it would skew the middle hands to the right of the distribution with an increased swelling in the 2-4 double range.

[HotelFSR]

Like I said 3 & 4 han become the most common. You can look at these kind of stats for various reach mahjong games/sites such as mahjong fight club and tenhou.net

The higher you go up the han scale the less you see red fives making an overall impact (because the larger hands are rarer to begin with before you factor in red fives).

Mangan score actually becomes marginally the most common if you consider that 7700/11200 (4-han) is essentially mangan.