Week 48: how to get better at chess endgames, one king move at a time
Week 48 was a quiet recovery: 12 rapid games from December 2 to 6, 2024, 7 wins, 4 losses and 1 draw, and my rating went from 1074 to 1099, with a high of 1107. My average accuracy, 82.8 %, was the best of this whole stretch of the diary. And yet the most instructive game of the week was a loss from a winning endgame, decided by where I put my king. So this week is about how to get better at chess endgames, starting with the piece I keep forgetting about.
The week in numbers
| What | Number |
|---|---|
| Games | 12 (#1157–#1168) |
| Wins / losses / draws | 7 / 4 / 1 |
| Average accuracy | 82.8 % |
| Blunders (my moves) | 29 (2.4 per game) |
| Rating | 1074 → 1099 (+25) |
| Most played openings | London System (2), Queens Pawn Opening Accelerated London System (1), Italian Game (1) |
With White I went 5–1, all with my 1.d4–Bf4 setup. My cleanest game was game 1163 against deborahshe (1081), 93.1 % with no blunders and no mistakes. With Black I went 2–3–1, and I was still trying new things: a first win in what Chess.com calls the "Modern Defense with 1 e4", in game 1164 against Cams1979 (1077), and a loss in the Modern against the Three Pawns Attack.
The endgame was where the week leaked. Seven of my 29 blunders came in endgames, and two games show why.
How to get better at chess endgames: the king walk in game 1158
With White against ItIsOkay2 (1079), in a London System, I reached an endgame with rook, knight and six pawns against rook, knight and three pawns. The engine said +5.63, "white is winning". Then I started moving my king.
Kg3 looks harmless. The engine's reply is Nh5+, "knight moves close to the enemy king: check", and in its line the knight keeps checking (31...Nh5+ 32.Kh3 Nf4+ 33.Kg4) while my advantage melts to about equal (+0.32 in the engine's line, +1.74 at the time). Ke3 also gets checked, 31...Nd5+ 32.Kd3 Nf4+ 33.Ke4, but there the king walks towards the centre and the engine keeps "white is winning" (+4.99).
What happened in the game is the real lesson: my three endgame blunders were all king moves, 31.Kg3, 34.Kh4 and 36.Kh3, each one further towards the h-file. After the third, the engine had me at -3.14, and I resigned a few moves later.
Two more moments from the week
Game 1157: trading my best bishop with tempo for Black
With White against Arbik121 (1071), in an Accelerated London, I was clearly better (+2.34). My bishop on e5 was a strong piece, and I traded it for a knight.
Position.py describes Bxf6 as a double attack: the bishop on g7 with the rook behind it, and the queen on d8. But Black takes back with the bishop, and that recapture "attacks the queen in the center", my queen on d4. The score went to -0.62. Keeping the bishop with f4 or Nc3 kept "clearly better for white". I still won the game by resignation.
Game 1159: the pinned knight
With Black against Hojiakbar1976 (1086), in an Italian Game, I was already worse (+2.60). White's bishop on g5 stood on the same diagonal as my knight on f6 and my queen on d8.
After Bg7 the engine's reply is e5, a pawn that "attacks the knight on the king's side", and in its line White wins the knight (8.e5 O-O 9.Bxf6 Bxf6 10.exf6): +5.87. Be7 blocks the diagonal; the engine still rates it "white is winning" (+3.01), but at least it's a game. I abandoned the game after 8.e5 Qe7 9.O-O.
The other endgame leak was my draw, game 1162 against jjpearce1986 (1069). I had 8 blunders in 116 half-moves; the last one, 51...Rxa5, took the engine's evaluation from -6.13 to 0.00, and the game ended king against king.
The lesson of the week
Good accuracy doesn't protect you in the endgame if the king is in the wrong place. In the middlegame I hide my king; in the endgame I keep treating it as something to hide, and walk it to the edge. The engine's preference in game 1158 was the opposite: bring the king towards the centre and towards my pawns, even if it gets checked on the way. So my endgame rule from week 48 is short: in an endgame, every king move should go towards the centre or towards the pawns, never towards the corner, unless the engine (or a concrete calculation) says otherwise.
All games this week
| Game # | Date | Colour | Opponent (rating) | Result | Opening | Accuracy | Link |
|---|---|---|---|---|---|---|---|
| 1157 | Dec 2 | White | Arbik121 (1071) | Win | Queens Pawn Opening Accelerated London System | 88.9 % | game |
| 1158 | Dec 2 | White | ItIsOkay2 (1079) | Loss | London System | 81.9 % | game |
| 1159 | Dec 2 | Black | Hojiakbar1976 (1086) | Loss | Italian Game | 79.3 % | game |
| 1160 | Dec 3 | Black | Rasakparuvingal (1087) | Win | Giuoco Piano Game | 81.6 % | game |
| 1161 | Dec 3 | White | Jovem56 (1046) | Win | Modern Defense with 1 d4 | 91.7 % | game |
| 1162 | Dec 5 | Black | jjpearce1986 (1069) | Draw | Indian Game East Indian London System | 80.8 % | game |
| 1163 | Dec 5 | White | deborahshe (1081) | Win | Indian Game | 93.1 % | game |
| 1164 | Dec 5 | Black | Cams1979 (1077) | Win | Modern Defense with 1 e4 | 89.2 % | game |
| 1165 | Dec 6 | White | GMlequangliem9B (1076) | Win | English Defense | 79.9 % | game |
| 1166 | Dec 6 | Black | shubhamxingh (1128) | Loss | Modern Defense Three Pawns Attack | 68.1 % | game |
| 1167 | Dec 6 | White | vikash_patori (1079) | Win | London System | 83.4 % | game |
| 1168 | Dec 6 | Black | 69Joe69Mama69 (1112) | Loss | Nimzowitsch Defense Declined | 75.9 % | game |
Previous: week 47 · Next: week 49 · The journey from 631 to 1385 · Review your own games like this
Frequently asked questions
How do I get better at chess endgames as a beginner?
My week 48 games point to king placement first. In game 1158 all three of my endgame blunders were king moves toward the edge of the board, where Black's knight could keep checking it; the engine's choice was a king move toward the centre.
Can good accuracy still lose games?
Yes. My average accuracy this week was 82.8 %, the best of this period, and I still lost 4 of 12 games. Game 1158 was played at 81.9 % and lost from +5.63.
What is a common endgame mistake at 1000–1100?
In my games, giving away a winning position with one careless capture: in game 1162 my 51...Rxa5 took the engine's evaluation from -6.13 to 0.00, and the game ended king against king.