A xiangqi endgame chart in plain words: which material wins, which needs exact technique and which is a dead draw, from a lone chariot to soldiers.
A xiangqi endgame chart answers one question: does the material left on the board win? A lone Chariot beats Advisors and Elephants once one of the four is gone, but draws against the full set. A Horse alone beats only a bare General or one Advisor, a Cannon needs an Advisor of its own, and Soldiers depend on height.
Throughout, Red is the side with the extra force and Black defends. Advisors and Elephants are the guards, and the full guard, 士象全, means both of each. Each balance below gets one of four labels. A win means the books list it as a book win, 例胜, which the stronger side converts from almost any start. A practical win means the stronger side wins if it reaches one key setup first. A hard win, 难胜, means correct defence draws and the attacker only wins against a badly placed guard. A draw means there is no winning method at all.
The verdicts come from the standard Chinese references: the Chinese Wikipedia summary of 象棋勝和定式, which follows Jin Qichang and Yang Dian's book of the same name and Xue Rongxian's beginner's guide, and Tu Jingming's Practical Endgames as presented on xqinenglish. Where they help, we add figures from Jih-tung Pai's endgame databases, which were built by retrograde analysis. Any mate or stalemate described here was replayed on the rules engine that runs XiangqiPlay.
A lone Chariot
Against a bare General: win. The Chariot hardly needs its own General. Put Black's General on a back corner of its palace and the Chariot on the palace centre, and Black has no legal move, because the Chariot covers both neighbouring points. In Xiangqi a player with no legal move loses, so that is the end of the game.
Against two Advisors, two Elephants, two Advisors and one Elephant (单缺象) or one Advisor and two Elephants (单缺士): win. Losing one guard is enough. Pai's database finds a win in every position of these four endings where the Chariot side is to move.
Against the full guard: hard win. If the General stays home and the guards stay low, central and linked, a lone Chariot cannot break in. It only wins against a scattered setup. We look at that fortress in detail in our Chariot versus full guard guide.
Against a lone Horse or a lone Cannon: win. Against a Horse with two Advisors: win, a classic that the books teach by driving the Horse to the side with no Advisor. Against a Horse or a Cannon backed by the full guard: draw in practice.
Against a Cannon with two Advisors: hard win, and placement decides it. With the Cannon tucked in front of its General and the Advisors linked, the Chariot cannot make progress. Against a Cannon with two Elephants, Tu Jingming says only two formations hold, and anything else gives the Chariot good winning chances. Against two Horses, two Cannons, or Horse and Cannon: hard win.
A Horse or Cannon on its own
Horse against a bare General: win, and the finish is a stalemate rather than a check. In the book line the Horse herds the General into a back corner, Red's General takes the central file, and one quiet waiting move leaves Black without a legal move.
Horse against one Advisor: win. The method wins the Advisor first and then repeats the stalemate. Horse against one Elephant: hard win, held when the Elephant and the General sit on opposite sides of the board, so the Horse cannot watch both. Horse against two Advisors: usually a draw. Against two Elephants or the full guard: draw. Against a Soldier that has crossed the river: draw.
Cannon against a bare General: draw. A Cannon needs a screen to attack anything, and the defender never has to give it one. We checked every placement of the two Generals and the Cannon with Black to move, and in none of them is Black left without a legal move, so a lone Cannon can never win, not even by stalemate.
Cannon plus its own Advisor against a bare General, one Advisor or two Advisors: win, because the Advisor becomes a permanent screen. Two Cannons against two Advisors: win. Two Cannons against two Elephants: draw, since the Elephants keep blocking the central file. Two Cannons with an Advisor of their own against the full guard: win.
Soldiers: high, low and old
Height matters more than anything for Soldiers. A high Soldier has crossed the river but has not gone below the defender's Soldier rank. A low Soldier is deeper, on the palace's front or middle rank. A bottom Soldier, also called an old Soldier (底兵, 老兵), has reached the back rank and can only shuffle sideways.
One Soldier against a bare General: win if it is high or low, draw if it is a bottom Soldier. The low-Soldier finish is another stalemate: the Soldier sits on the palace centre, the General is stuck on a back corner, and a waiting move by Red's General leaves Black with no move. A bottom Soldier never gets there. We checked every placement of a back-rank Soldier and both Generals, and Black always has a legal move.
One Soldier against one Advisor or one Elephant: usually a draw, with wins only from awkward placement.
Two Soldiers, one high and one low, against two Advisors: win. Against two Elephants: win, as long as neither Soldier is on the bottom rank. Two low Soldiers against two Advisors: hard win. Two Soldiers against an Advisor and an Elephant: hard win. Against the full guard: hard win at best, and Pai's database finds the defence holding in nearly every position. Three high Soldiers against the full guard: win.
The neatest lesson on height comes from a lone Cannon. Two high Soldiers beat it, because they can still choose their road. A high and a low Soldier only draw, because the Cannon parks on the central file and the low Soldier cannot go back to look for another road.
Pairs against the full guard
Chariot and Soldier against the full guard: win. Pai's database finds the Chariot side winning in practically every position with it to move, and Tu Jingming shows a Chariot with a high Soldier beating even Horse and full guard. That is why it pays to keep one Soldier before you trade down. Against Cannon and full guard the same pair is only a hard win.
Two Horses against the full guard: win. Two Cannons without an Advisor: usually a draw. Cannon and Soldier: draw, and Pai's database finds only about one position in twenty won. Horse and Soldier: usually a draw, with the details in our Horse and Soldier guide.
Chariot and Cannon against a lone Chariot: practical win. Whoever's Chariot holds the central file decides it. Chariot and Horse against a lone Chariot: practical win, which Pai's database scores as a win in roughly nine positions out of ten. Give the defender guards and both become harder, which is the subject of a separate guide.
Using the chart in a real game
The chart earns its keep before the endgame starts. If you are the side with the Chariot and the opponent still has all four guards, do not swap off your last Soldier, because Chariot against the full guard is a draw with correct defence and Chariot and Soldier is a win. If you are defending, count your guards before every trade and refuse any line that costs one for nothing.
When two results look close, prefer the one where the defender has fewer drawing setups to remember. A hard win against a nervous defender is still worth playing on, but you should know that correct play holds it.
These balances stick far better once you have played them. Set a few up on a board and play both sides, then watch for them in your own games against XiangqiPlay's computer, which runs at eight levels from Easy to God and needs only a free account.
XiangqiPlay Team
XiangqiPlay newsroom
