mahjong's push-fold and defense | Explaining the judgment of genbutsu, suji, kabe, and safe tiles

The safety level is determined not by “each tile” but by “each opponent”.

Even in the same case, the upper family may have genbutsu, the opposite person may have suji, and the lower family may have no clue. Safety is not a fixed property of tiles. Decide first who you don't want to deal-in with, and make a judgment based on the opponent's river, vice, riichi declaration, and visible tiles.

When riichi enters from two people, even if you cut off the genbutsu of only one, there is a possibility of a deal-in to the other. The common genbutsu in both rivers is the first candidate. If there is no common genbutsu, divide the grounds for each opponent into two columns. If you specify the target, such as ``genbutsu for upper house, suji based on genbutsu of 4000 for face-to-face'', you can prevent misunderstandings in safety evaluation.

The same mindset is required for vice-representatives. The wait is not necessarily fixed like the riichi's, and the shape may change depending on the hand. Although you can read candidates for visible yaku such as ``dye'', yakuhai, and taiwa, you cannot determine the invisible hand. Let's consider the estimation of risk and the probability of hitting in a deal-in separately.

genbutsu: the same tile that the target opponent has already discarded

genbutsuis a tile that the opponent you want to protect is throwing away into your own river. According to furiten rules, that opponent cannot ron with the same tile that you discarded. If you want to completely get off on a riichi person, the easiest procedure to explain is to first look for genbutsu. It doesn't prevent you from win by tsumo, but you can avoid rondeal-in to that opponent with your own discard.

Importantly, genbutsuonly to that personIt's about passing. Even if the upper family abandons Shiman, if the waiting time for the meeting is Shiman, then the meeting will be successful. Beginners should first practice picking up the tiles in the target's own river as genbutsu to avoid confusion.

When you miss another player's discarded tile, furiten has different lengths before and after riichi. If a non-riichi player misses a winning tile, it will become a furiten within the same round, but it will be resolved after completing his/her next move. If you miss the winning tile after riichi, furiten continues until the end of the game, and after that you cannot win and only win by tsumo is valid.

Therefore, the same tile that the other player cuts after riichi and the riichi player does not roll will pass against that riichi player until the end of the game. On the other hand, even if the same tile passes once to a non-riichi player, the hand or wait may change after the next stroke, so it is not considered safe until after that. Both have different bases from the target's own river, so check the order in which they were cut and the other party's riichi status.

There are also aspects to the order in which you cut genbutsu. If you have multiple genbutsu, compare the tiles that are likely to remain as safe tiles after the next round, the tiles that can be used by another attacker, and the tiles that are unlikely to prevent your hand from reviving. Instead of thinking, ``I'm going to cut everything right away,'' I think ahead to what to cut next.

suji: Clues to exclude combinations with double-sided waiting from the river

sujiThe idea is to use the furiten and sequence structures to exclude certain double-sided waits. Number tiles can be organized into three systems: 1-4-7, 2-5-8, and 3-6-9. For example, if 40,000 is genbutsu, the opponent cannot ron with 40,000 even if he is waiting for 1/40, 230,000, or 4/70,000 of 560,000. Therefore, the risk of 10,000 and 70,000 is lowered compared to waiting for both sides.

However, this does not guarantee that if 40,000 is safe, 10,000 and 70,000 are also safe. Ichiman single rider, Ichiman and honor tiles' combination, and other hand-made combinations remain. Suji only explains part of waiting for both sides. Remember that the base is narrower than genbutsu, and match it with the number of visible cards and the opponent's hand.

1-4-7 suji

2-5-8 suji

3-6-9 suji

When looking at suji, we distinguish between thinking about both sides from the central genbutsu and thinking about only one side from the edge genbutsu. For example, even if 10,000 is genbutsu, there will still be 56,000 waiting for 4 and 70,000, so the entire 40,000 cannot be said to be safe. In addition to memorizing the numbers as three columns, you will also see which double-sided tarts disappear with a hand example.

Waiting for single horse/souji/mebari that cannot be prevented by suji

suji is a guess based on waiting for both sides. don't use sequenceWaiting for solo riderandWaiting for twins, wait for the one in betweenWaiting for fittingThis does not apply directly to . Let's look at the typical reasons for deal-in to suji tiles.

10,000 single horse: Even if 40,000 is genbutsu, 10,000 can win

Ichiman and Higashi no Sohoku: Waiting for two types without using sequence

Waiting for 13 cylinders and 2 cylinders: 5 cylinders will not disappear.

In addition, there are yakuhai's single horse, seven-pair child's single horse, and complicated waits that arise from rearranging sequences. Therefore, suji is not a ``safe tile'' but a ``candidate that is unlikely to hit part of the double-sided waiting list''. If you want to get off completely and use genbutsu, this is why you should prioritize genbutsu over suji.

Kabe, No Chance, One Chance: Narrow down the sequence by looking at 4 images

kabeThe idea is to take advantage of the situation where all four tiles of the same number are visible to you. If you can confirm four hand, river, sidero, and dora display tiles, your opponent cannot hide those tiles and use them in the sequence. For example, if you see 4 30,000, your opponent cannot have 230,000, so you cannot wait for 10,000 on both sides.

In this way, when the four necessary constituent tiles are visible, the possibility that a certain tile will be used for a specific double-sided wait has disappeared.no chanceIt is called. When Ichiman sees 4 Sanman, he has no chance against 23,000 waiting for 1 and 40,000. However, there will still be a waiting list for Ichiman and Ichiman, so I cannot say that it is ``absolutely safe'' here either.

one chanceis a state in which the three necessary constituent tiles are visible. If your opponent uses the last card you can't see, you can create a sequence wait, so the basis is weaker than no chance. Furthermore, it is not known whether the one piece that is not visible is in the pile or whether it is in someone else's hand. It can be used as a comparison material when the round progresses and there are no more tiles to cut, but it cannot be treated the same as genbutsu.

Kabe differs from suji in that its hand can also be included in the information. For example, if you have 30,000 cards in the dark and you see one card remaining in the river, you have four cards. On the other hand, if you count the wrong tiles, the conclusion will be reversed. Rather than checking whether you can see the 4 Ichiman you are trying to cut, check how you can see the 20,000 or 30,000 that will prevent your opponent from making the shape of 23,000.

Since honor tiles do not have suji on both sides, it is necessary to clarify which three tiles "3 tiles visible" refers to. If you already see 3 of the same honor tiles outside your hand and you have a 4th one, your opponent cannot have any tiles of the same type. The fourth card does not correspond to the normal single horse, seven pairs of children, or yakuhai triplet or double check. However, there is still a possibility that Kokushi Musou who can wait without having any of the target honor tiles will be defeated.

On the other hand, if "3 cards visible" includes the one card you are about to cut, and only two cards are visible outside of your hand, one card is still missing. If your opponent has that one card, you can not only wait for a normal single horseman or seven pairs of warriors, but also wait for Kokushi Musou. When counting the number of visible pieces, distinguish between whether it is the total number that includes candidates to cut, or whether it is the number of visible pieces outside of your hand. Do not put the number tiles and honor tiles in the same suji table, and evaluate based on the estimated number of remaining tiles.

Push-fold decision flow for beginners

Even if you have knowledge of defense, it doesn't mean you can get off every time. Your win is worth the score and the game, and if you are a parent, winning is also important. On the other hand, if you pass too many dangerous tiles from far and cheap hands, the chances of losing points will outweigh the rewards. By verbalizing the conditions in the following order, you can reduce judgments based solely on feeling.

1. Who is attacking and how much:Check riichi, mechanism, how dora looks, and parent and child. If multiple people are attacking, assign a safety level to each.
2. What I am:The number of danger tiles required to win is different for hearing tiles, 1shanten, and 2shanten or less. The farther your hand is, the harder the conditions for holding down will be.
3. Are RBIs and ranking worth enough?Yaku, Dora, parent's consecutive matches, opponents you want to hit directly, see if the order of finishing changes if you win. If the number of han is not determinedriichi mahjong's yaku listandScore calculation guideYou can organize it with
4. If it's a hearing card, look at the quality of waiting:The likelihood of winning differs depending on whether you can expect to have many cards left on both sides or if you only have one card left. Also check if there is a final stage where formal hearing tiles have value.
5. How many danger tiles need to be passed:In addition to the current one, think about whether you need more dangerous tiles before your next move or win. You don't have to press it all the way once you start pressing it.
6. Check the round, number of remaining games, and point difference:In the early and late stages, the first game in the east and the final hand, and the top and last hands, the value of defending and the value of winning change even with the same hand.
7. If you want to get off, choose from genbutsu:After deciding on a strategy, return to searching for safe tiles. Compare genbutsu, common genbutsu for multiple people, kabe/no chance, suji, etc. by target audience.

In this flow, there are no universal numbers that apply to everyone, such as ``At what percentage of the deal-in rate should I press'' or ``At what round should I always get off?'' This is because the profit or loss changes depending on the recruitment rules, starting points, ranking points, red fives, the situation, the opponent's batting points, and your waiting. When using numerical values, check the population and assumptions for the numerical value, and use it as a reference only in situations where the conditions are similar.

Three examples where the same dangerous tile can lead to different conclusions

Example 1: A well-shaped mankancho tile with only one tile to pass through.

Situations in which you can maintain your winnings by cutting one unknown tile against your opponent's child riichi with two-sided full-sided tiles, the value of winning and the quality of waiting will be a factor. However, if your opponent is a parent, the chances of losing points will be greater, and if you are the top player in the final hand, the value of defending your position will be added. Instead of "always press the tile because it's a Mankancho tile", line up the changed conditions horizontally.

Example 2: Endgame with cheap 2shanten

With 2shanten only for yakuhai, there are few rounds left and if you have to cut multiple dangerous tiles in the future, the distance to win is a big weakness. If you have a safety tile, it will be easier to think about the direction in which to get off. How to compare the effective tiles of handmahjong's introduction to tile efficiencyAs you can see, there is a reason to prioritize safety in the final stage, not just effective tiles.

Example 3: Win is the final hand condition

In the last position of the final hand, if you win, you will move up in the rankings, and if you lose, you will be closer to the end, so the ranking conditions are the reason to attack. On the other hand, if you make a deal-in at the top and the odds are reversed, the value of avoiding points will increase even with the same hand. In order to avoid ambiguity regarding profit and loss and finish conditions,Settlement rules including uma and okaIt is also important to share.

Read the M League official results by arranging multiple indicators side by side.

The M League official website publishes the total number of games, average finishing order, score rate, deal-in rate, riichi rate, secondary rate, etc. for each player. The following table is2025-26 regular season display confirmed on official results page on July 25, 2026is part of. Decimals in published values ​​are replaced with percentages.

Even with these three lines alone, the player with the lowest deal-in rate and the player with the lowest average finishing order do not match. However, it is not possible to simply decide that one is better, defensive or offensive. Differences in styles and results can only be considered by combining the attack rate, riichi rate, secondary rate, RBI, number of matches, and selection for each phase.

Also, this is not a table that ranks players by a single metric, but an example of reading multiple metrics at the same time. If the period changes, the sample size and numbers will also change. M League official results are live public data and are updated depending on the season and time of display, so when using the latest values,Official results pagePlease check again.

Estimating waiting tiles and push-fold judgment are not the same problem

In public research, a method has been proposed to estimate waiting tiles by considering the hand configuration of the riichi player. This is an incomplete information game in which mahjong cannot completely see the other player's hand, and the problem is that prediction is important to avoid losing points due to discarded tiles. genbutsu, suji, and kabe can also be considered clues for humans to narrow down candidates from public information.

On the other hand, even if you can estimate the waiting tile, whether you should press it or not is a different matter. You must compare your win probability and points scored, points conceded during deal-in to your opponent, changes in finish order, and number of remaining games. A push-fold conclusion cannot be drawn just by reading the waiting risk level. This is the basis for defense and the reason for dividing the value judgment of the entire situation into two stages.

Research on mahjong analysis using tiles shows the idea of statistically handling a large number of match records and separately considering individual discard techniques and strategies related to overall performance. It is important not to judge the quality of a decision based solely on one deal-in, but to review the choices made under the same conditions. Even if you play a safe tile, your opponent may still win, and you may win by pushing a dangerous tile. Separately record the results and decision process.

Practice to acquire defense and push-fold

The first exercise is to focus on one person's river and annotate each tile in the hand with ``genbutsu'', ``suji'', ``shikabe'', and ``no basis''. The goal is to avoid making mistakes in classification before predicting the correct winning tile. Once you get used to it, the two of you, riichi, separate the line facing Kamie and the one facing you, and look for a common genbutsu.

Next, record the stations you lost during push-fold using six items. My number of shanten, expected RBI, wait, parent and child, round, and point difference. Finally, write down the classification of the tiles you cut for each opponent. Rather than thinking it was a mistake because you made a deal-in, review whether the information available before the deal was balanced against the value you were trying to get.

In the aggregation, check not only the deal-in rate but also the deal-in rate, average order of arrival, first-place rate, and last-place rate in the same period. Even if the deal-in rate decreases, if the winning rate drops significantly and the finish order does not improve, you can consider the possibility that the deal-in rate has dropped too much. On the other hand, if the last-place rate is worsening even though the scoring rate is high, look back at the ways in which you conceded a lot of goals and the score situation. It is important to return to the memo of the situation and not conclude the cause-and-effect relationship with just one number.

Common defensive mistakes

• Make one genbutsu a safety tile for everyone:genbutsu differs depending on the target audience. Attacks by multiple people require a common source or basis for each opponent.
• Call suji absolutely safe:suji excludes certain double-sided waits. Single horsemen, soukoku, kabari, etc. will remain.
• Confusing one chance with no chance:If you see 3 cards, your opponent may be able to use the last one. The strength of the evidence is different from seeing 4 images.
• Do not include your hand in the visible number:Kabe counts not only the river, but also all the information that can be seen from the player, including the hand, sideroof, and dora display tiles.
• Press just to hear the tile:You can check the number of cards remaining in the waiting list, the score, the number of dangerous tiles, whether the opponent is a parent, and the order of arrival conditions.
• I don't think you can get off once you press it:Conditions will change with the addition of new riichi, danger tiles, and reduction of waiting tiles. The judgment will be updated every round.
• Decide your strategy based on a single result:The basis for judgment regarding dangerous tiles that were not dealt-in remains weak. Separate results and selection information.

Summary: Separately write down the basis for safety and the value of promoting it.

In defense, first separate the opponents one by one and search for that opponent's genbutsu. If you don't have genbutsu, use Kabe/No Chance, suji, honor tiles, etc., where you can see 4 tiles. Suji is a clue to organize double-sided waiting for 1-4-7, 2-5-8, and 3-6-9, but it is not a safety guarantee because it does not apply to single horse, sogo, and kamenbari.

With push-fold, you can check your Shanten number, RBI, waiting list, parent/child, rounds, and score status. The attacks of multiple people, the number of dangerous tiles passed, and the number of remaining games are also added. There is no numerical threshold that applies in every situation, and the conclusion will change if the conditions change. To improve your decision-making, don't just track the deal-in rate; instead, continually reflect on the basis for your choices, along with your last-place rate, deal-in rate, and match history.