Lately, I've been playing "Jantama" (麻魂) online mahjong instead of "Tenhou" (天鳳).
I mainly play in the "Jade Room," and since my performance has stabilized and I've reached Mahjong Master ★3, I'd like to write about my current situation, including how to play in the Jade Room (as of January 2021).
Mahjong strategy book"The Impact of Suphx, the World's Strongest Mahjong AI"I feel that many of the ideas I've adopted after reading it are proving very useful.
The Impact of Suphx, the World's Strongest Mahjong AI (Mynavi Mahjong BOOKS)
Current performance
First, I'd like to do a self-analysis of my current academic performance.
Rank distribution:
I'm personally satisfied because the rankings from 1st to 4th place are in an inverted pyramid shape.
I consider myself to be somewhat defensive, but if I play too defensively, I tend to get fewer first and fourth place finishes and more second and third place finishes.
However, I'm happy that we've managed to get first place quite a lot so far.
I feel like I'm able to behave appropriately according to my points and finishing position.
Average rank: 2.31st
Looking at the records of the top-ranked players in Mahjong Soul, there are some with incredible rankings like 2.27th, but I'm not at that level.
Even with the current ranking distribution, my point balance is positive, so I think I'll naturally get promoted if I keep playing matches at this pace.
My performance worsened for a while when I started hitting the balls too close together, but it's improving now.
For players who have made it through the Gold Room, their average ranking will naturally worsen in the Jade Room, where the opponents are stronger.
It's incredibly stressful when your previously impressive performance starts to decline, but until your results stabilize to some extent, just accept it as data collection and keep going at it calmly.
Average sum: 6197
This is the average score when winning a hand.
This number is quite high; even among the top-ranked players, you won't find anyone with a higher number.
I am certainly quite conscious of my RBIs, but I was surprised to find out that this number is among the top when I looked it up.
However, this isn't simply something to be unconditionally happy about, as there are things we've lost as a result (more on this later).
Number of winning rounds: 12.12
The top-ranked players are mostly those with a score below 12.00.
This number represents the average number of turns it takes to win, so I think it simply means that I'm inferior to the top players in terms of tile efficiency and other factors.
I suspect that the reason their average score is high is because they're sacrificing the number of turns it takes to win.
Since you rarely see players who focus on scoring points at the top level, it seems that prioritizing winning rate and number of turns to win over scoring points is more likely to lead to a higher ranking.
It makes sense that a move with a low, fast point is stronger than a move with a high, slow point.
Win rate: 22.72%
やや低めの数値だと思いますが、和了率は常に放銃率とのセットで見ないと意味がありません。
I think these numbers come about because I'm a defensive (self-proclaimed) and somewhat timid type of player.
Immediately after reaching the ball area, I had around 25%, but the fact that it has dropped this much suggests that more of the surrounding players are also folding properly.
放銃率：10.81%
I think I'm doing a pretty good job of maintaining those numbers.
トップ層は「和了率－放銃率」の値が12～13なので、12を目指して打っていければと思っています。
この「和了率ー放銃率」の値は、野球でいうOPSのように、プレイヤーの実力をざっくりお手軽に示す指標だと感じています。
A score of 9-10 suggests that the underlying strength is still insufficient.
Secondary dew rate: 33.68%
Personally, I feel the numbers are pretty much in line with my intuition.
However, in order to improve the number of turns it takes to win and the winning rate, and to aim for further improvement, I think it will be necessary to call tiles up to around 35% of the time.
I've gained a fair amount of confidence in being able to fold after making a sound, but I often find myself thinking, "Ah, I should have made that sound."
If you can improve this aspect, your average score will decrease, and your winning rate and number of turns to win will likely improve.
The likely cause is that players are a little late in calling tiles, likely because they dislike the idea of lowering their score and having to deal with a hand that only partially completes.
早いアガリは絶対的な正義なので、ここを上げつつ放銃率の上昇を抑えるというのは、今後の課題のひとつです。
Recovery rate: 19.76%
While your riichi rate might be slightly high, it's neither extremely low nor extremely high, so I think you don't need to worry about it for now.
If you increase the rate of open hands, the number of open hands will naturally decrease.
I think my score has dropped slightly because I've become able to play more calmly compared to the time when I was just calling "reach" without thinking.
For Those Who Always Finish 2nd or 3rd in Mahjong | Understanding Average Rankings and Why You Can’t Seem to Win
Things I've started to become aware of and the results I've seen
Things to keep in mind in the early stages
If you have few potential melds at the start of your hand and your potential score is likely to be low, you will have fewer options to push forward when you are on the defensive.
In cases of such "bad" starting hands, I believe it's important to focus on building a good hand or accepting dora tiles, even if it means taking a slightly longer route.
Ultimately, if another player scores first, you can't push back unless you score a certain number of points yourself.
If the starting hand is bad and the game is going to be slow anyway, the intention is to at least secure points and build a hand that can push back, thereby increasing the chances of winning.
This way of thinking probably connects to my tendency to "score high points + win later in the game."
While it's true that winning quickly is advantageous in mahjong, I feel that it's better to steadily build up your hand without calling tiles when it inevitably slows down, as this provides a more stable defense.
It's also important to consider whether the move is something that can be played in front of the opponent.
If you have a lot of bad shapes and it looks like you'll have to rely on calling tiles anyway, you'll want to cherish isolated honor tiles that increase your score just by stacking them more than isolated 1s and 9s, and try to hold onto them a bit longer.
・Specific examples in actual combat
With this starting hand, it seems difficult to keep the hand concealed, so it appears that the best option is to call "pon" on the white tiles.
Therefore, even if you call, you should aim to secure some points by holding onto the East tile and the 7p next to the dora tile for as long as possible, while also keeping an eye on the possibility of a Chanta hand, and at least aiming for around 2000 points.
The two North tiles, which are both potential pairs and safe tiles, will be kept safe without calling a pon.
Given the current score situation, there's no need to stake everything on this hand, so it's important to have an escape route in place so you can fold even if you call a tile.
With this hand, you can expect either a closed hand or a hand with one dora tile.
However, since getting double East or Dora Middle can be very big, you'll want to discard 9p or 9s until you have a reasonable chance of winning with a closed hand.
Since there are a fair number of kanchans, if you get any more honor tiles, you should call them as much as possible, and proceed with a balance between playing with a closed hand and playing with open tiles.
This is a hand with four dora tiles, so it's a hand where you want to aggressively call tiles and push forward, whether it's for a simple hand or anything else.
However, the leader is far ahead, making a comeback quite difficult, so you'll want to hold onto some safe tiles that allow you to fold if the last-place player (the dealer) makes a move.
You can push if you have a good hand if everyone's scores are even.
In situations like this, where it's difficult to aim for first place and there are strong contenders for last place, I think it's necessary to be cautious about going all-in just because a mangan or higher is guaranteed.
Things to keep in mind in the midfield
Even if no other player has taken the lead by the time Kawa enters the second row,If the opponent gains the upper hand, you need to check whether you are in a position to press forward.
Considering the score situation, let's think about whether it's really necessary to push hard in this round.
If you are in fourth place in South 3 and the point difference between you and third place is more than 10,000 points, you absolutely must close the gap in this round, otherwise the last round will be extremely difficult as you will need to reach a haneman (a high-scoring hand).
Conversely, if you're not in that situation, and you have more than the starting score or more than 30,000 points, there's less need to press the button.
You probably don't want to give up the top spot so easily, but if you make a mistake here, you could end up not just second place, but third place or even last place.
I feel that it's easier to accumulate points by making moves that ensure you avoid coming in fourth place, rather than taking risks to try and get first place.
For example, in a situation like this, discarding the 5s is a preemptive discard to secure a safe tile.
Since it's still early in the game, it's fine to discard the 3p here, but depending on the density of the other players' discards, you might want to discard it earlier.
Since the score will be low anyway without a Pinfu, it's fine to discard the 5s first from the 7th turn onwards.
The red 5s has already been discarded, and the lack of appeal in a concealed triplet of 5s is another reason to encourage early discarding.
Unless there's a very good reason, if you're one step away from a winning hand and another player has already made the first move, you should avoid pushing a risky move.
Therefore, it becomes quite important to discard tiles early to fix a two-sided sequence and hold onto safe tiles, rather than maintaining a perfect one-shanten hand.
Of course, in terms of tile efficiency, a perfect one-shanten hand is advantageous, but having one safe tile available when another player declares tenpai is extremely important.
This approach is a distinctive one that is also incorporated into the Mahjong AI "Suphx," and it seems to involve many decisions that involve balancing things to prepare for future risks.
It seems they value the risk of "discarding a high-risk tile" as a significant negative, and I personally like this way of thinking because it suits my own style.
和了率が低くなり放銃率も下がる守備的な考え方なので、現在のスタイルと大きく異なる人は、実践するのに抵抗があるかもしれません。
However, Suphx often pushes for a hand that's one step away from a ready hand in the early stages of a riichi, so judging when to use each option is difficult.
Things to keep in mind in the final stages
If the river is in the third row, then we should probably abandon the optimistic assumption that all the other players are not in a winning hand.
Your primary focus should be on avoiding making a mistake, and you should proceed as if you were folding against a declared reach.
Even if you have a winning hand, declaring "reach" isn't worthwhile unless you have a hand worth a decent amount of points.
At this point, you should pay attention to the tiles that other players discard.
A player who can play at the level of a professional mahjong player should be careful about which tile they discard, even if other players haven't given any explicit signs of intent to discard.
However, even so, players who are discarding quite a few dangerous tiles are highly likely to be in a ready hand.
Since it's the end of the game, many players won't declare a reach, so be careful of surprise attacks.
Also, if you can achieve a formal tenpai (ready hand) without discarding dangerous tiles, try to aim for it as much as possible.
If you have to cut the tendons after the animal has cried out, it's probably best not to force it.
Since calling "pon" consumes two safe tiles, it's safer not to call "pon" if you don't have a tile that is guaranteed to pass.
If you can see where the dora is and are confident that the other players' scores are low, you might be able to push a little, but generally, the priority is to play it safe and fold.
Things to keep in mind when you are on the defensive
Now, let's review what to do when another player declares a reach and decide whether to push or fold.
It's human nature to be reluctant to fold so easily when you have a good hand.
However, you need to hold back and roughly calculate the reasons for pushing and pulling before making a decision.
Below, I will list some of the main things to consider when pushing or pulling.
I'm not calculating and making decisions based on these exact points; please just consider this as an estimate of the magnitude of the impact I'm seeing.
I believe that subtracting is advantageous when the balance is zero.
<Elements to press>
- Two-sided wait + expected score of 2000 or more: +2, Two-sided wait + expected score of 7700 or more: +3
- Bad shape + expected to have a score of 5200 or more +1
- Only one other player has declared Riichi, and the waiting tile is likely to be discarded from the sidelines (honor tile, suji hook 19) +1
- If you finish in 4th place: +1, if you finish in 3rd place: +1
- There are absolutely no safe tiles +1
<Elements to subtract>
-1 shanten -3, 2 shanten -6
- Two players have declared Riichi -3
- If the attacking player is in 4th place -3, if they are in 3rd place -1
- If you come in 1st place: -2, If you come in 2nd place: -1
- When you have a ready hand, you must discard 4, 5, or 6, which are not safe tiles. -2
- When you have a ready hand, you must discard 2, 3, 7, or 8, which are not safe tiles. -1
- The number of tiles you are waiting for is less than half (4 or fewer for a two-sided sequence, 2 or fewer for a bad shape) -3
- A draw is expected in the endgame (third stage) -1
The image is roughly as described above.
As you can see, this decision is made when you're on the defensive, so there are some pretty tough conditions that must be met before you can push forward.
In particular, I want you to keep in mind that pushing when you're just one step away from a winning hand is extremely difficult.
I've factored in a generous amount of subtractive elements, so you'll understand when you see it.
While it's possible to discard tiles if they are actual tiles, 19 tiles in a sequence, or honor tiles, the important question is, "After taking a risk to reach tenpai, will it ultimately become a hand that can be played?"
When the balance is zero, subtracting is generally considered the more advantageous option.
In other words, even if you have a two-sided wait for 7700 points and are one step away from a winning hand, if you can't maintain it without discarding a dangerous tile, it's generally more advantageous to fold.
Even if you don't see many red tiles or dora tiles, it's important to consider the risk, as there's a high probability that other players have them.
If you make the first move with this reach, it's roughly equivalent to a +5 advantage, so you don't need to overthink it; often, reaching is the correct move.
However, if the parent is in fourth place, they often push back and overcome the pressure of being chased, so it's important to know if you can compete if you are being chased.
The "bad shape" is at a significant disadvantage when it comes down to a head-to-head exchange.Therefore, you should keep in mind that being chased could be dangerous.
In conclusion
I shared an analysis of my current performance in "Mahjong Soul" and discussed my strategies for pushing and folding.
Regarding the content of the judgment, I read it most recently."The Impact of Suphx, the World's Strongest Mahjong AI"It is strongly influenced by this.
The Impact of Suphx, the World's Strongest Mahjong AI (Mynavi Mahjong BOOKS)
Regarding changes in point evaluation based on your own points and the progress of the game,"Statistical Mahjong Tactics"That content has stuck with me.
The discussion about how the value of a mangan (full house) hand is completely different in East 1 compared to South 4, and how the value of a mangan is completely different when you have 15,000 points compared to 35,000 points, was quite practical.
Mahjong Tactics Based on Statistics (Modern Mahjong Tactics Series)
And for overall playing style, including initial hand placement and push/pull decisions, please see YouTube.「Unimaru Mahjong Channel」This is very helpful.
As a comprehensive practical video, you can learn what to consider while playing while having fun.