Defense is not folding: when to push and when to bail

1. Folding and defending are different instructions

Folding is a clean decision: break the hand, order every discard by safety, and accept that you will not win this deal. Defending is broader. You still keep a route to a win, but you stop paying any price for it. The difference is whether the hand still has conditions.

Suppose an opponent riichi and you hold a confirmed 2-han hand plus two genbutsu. Folding means discarding those two tiles and then dismantling a ryanmen if nothing safe arrives. Defending means playing the genbutsu first, watching two draws, and only then deciding whether the hand still deserves to live. One closes the door; the other leaves a window open.

Mixing the two words produces two familiar mistakes. One player folds for two turns and shreds a hand that only needed patience. Another has no winning value left but tells himself he is “defending” while continuing to fire middle tiles into a riichi.

2. Score first, hand second

The same unsuited 5p is a different tile in East 1 at flat scores and in South 4 when you trail by 12,000. In the first case a deal-in costs you a hand; in the second it can end the match. So the order is fixed: confirm the objective of the deal, then read the hand.

When you lead, the objective is usually to protect placement. You do not need to trade an 8,000-point risk for a 1,000-point win. Keeping safe tiles is worth more than one extra draw. When you trail, the objective becomes chasing points — but chasing does not mean pushing every hand. It means raising risk only when a win crosses a placement line or brings the gap back into reach.

Flat scores are the easy case: value and danger decide, with no placement adjustment. The hard case is the middle — three thousand ahead with two hands left in the South round. Use one simple test: if I deal in here, do I fall from contending for first to protecting third? If yes, bail. If no, let the hand decide.

3. Hand value against the cost of safety

Hand value is not just han. It is what the win changes. A 1,000-point hand is a small gain at flat scores, but it can decide placement when you trail by 1,500 in South 4. Conversely, an 8,000-point hand waiting on dead tiles is worth less than its score sheet suggests.

The cost of safety has three parts: the chance that a tile deals in, the loss if it does, and the future value you destroy by breaking the hand. Most players calculate the first two and forget the third, so they turn a hand that could still pivot into a pile of disconnected tiles.

Expected value can be stated without a spreadsheet: push EV = chance of winning × value of the win − chance of dealing in × loss − chance the opponent wins anyway × placement cost. You will not compute exact numbers at the table, but you should walk through all four terms. Counting only your own hand size is half an EV calculation.

A practical substitute is the safe-tile count. With four or more safe tiles, you can defend for two turns and keep options. With one or two, decide now whether to push or fold completely. Do not wait until every safe tile is gone and call the resulting panic a decision.

4. Riichi, calls and visible danger

Against a riichi, read three things: the declaration tile, the discards after riichi, and the suji the player has already passed. Tiles near the declaration are usually more dangerous. A sudden switch from manzu to pinzu after riichi often means a wait change, or a hand that was always honitsu.

Against calls, the danger is not “they called”. It is what they called and what they still need. Repeated calls in one suit raise the chance of honitsu or chinitsu. A player who calls a middle tile and then starts discarding terminals may already be waiting. Treat their discard order as a timeline, not a set of isolated tiles.

Visible danger also includes the dora indicator and dora already on the table. Middle tiles near dora, and the suit an opponent has carefully preserved, carry a surcharge. Before every discard ask: who benefits most from this tile? If the answer is an opponent, it is not “probably fine”.

Turn count is the last variable. Before turn six, opponents are less likely to be tenpai, so continuing the hand is reasonable. After turn ten, treat every unsafe middle tile as a possible deal-in. That does not ban pushing late; it means a late push must clear a higher bar.

5. Three rules and a 15-minute drill

Rule one: below 2 han, against a riichi, after turn ten, do not discard an unsafe middle tile. Play genbutsu, then suji, and break pairs if you must. The only exception is a win that directly changes placement, with a concrete source of points.

Rule two: with two or more called hands against you and no tenpai of your own, stop attacking when you have fewer than three safe tiles. This is not pessimism. It is accepting that the hand no longer has the resources to fight. Ranking the rest of your tiles by safety is more useful than hoping for one more draw.

Rule three: when leading, do not risk 8,000 points for a 1,000-point win. The value of a lead is that opponents must chase you. You do not need to prove anything with a high-risk push.

The drill: collect ten positions where an opponent has riichi. For each one, do only three things — write your hand value (how many han, whether a win changes placement), count your safe tiles, and decide push or bail within ten seconds. Afterwards, ignore the result and check whether your reasons were consistent. Do this for a week and you will see your default bias: too conservative, or too attached to a pretty hand.