How to read a chess opening

What an opening is

A chess game always starts from the same picture. Sixteen pieces a side, every one of them on a fixed square. So the first move and the one after it have been played over and over for centuries, and the groups that came up often were given names. Such a group is an opening.

Many names come from people. The Ruy Lopez is a sixteenth century Spanish priest, and Alekhine Defense is a player. Others come from places, as in Sicilian Defense, French Defense and Dutch Defense. The name itself carries no verdict. It only records who played it first and where it caught on.

An opening usually branches within a few moves. The Sicilian starts when White pushes the king pawn two squares and Black answers on the c file, and what White plays next splits it into dozens of paths. Each path is a variation, and variations get names of their own. That is why an opening name is normally a family name followed by a variation name.

Memorising openings does not make anyone strong at chess. It does help to know where you are standing, because then the next move is easier to choose. The name is what you call that place.

Names and ECO codes

Openings carry a code as well as a name. The classification is called ECO, and it is one letter followed by two digits. It runs from A00 to E99, five hundred codes in all.

The letter names the volume. Volume A covers games where White opens with neither the king pawn nor the queen pawn moving two squares. Volumes B and C cover the king pawn openings, split by whether Black answers symmetrically or heads for the French and its relatives. Volumes D and E cover the queen pawn openings, with the Indian systems where Black brings out the king knight and prepares the king bishop taken into E and the rest into D.

The digits are the position inside the volume. Nearby numbers are often related lines, but that is a tendency rather than a rule. The variations of the Sicilian, for example, stretch from B20 all the way to B99.

One family often spreads across many codes, and one code often holds several families. So this site keeps both doors. There is a way in by family and a way in by code, and each code page lists the families it touches.

The code is only a filing system. A lower code is not a better opening, and a higher one is not a rarer one.

The rules of this site

This section describes what the code actually does. The data comes from one licensed repository. Five files holding an opening name, an ECO code and a move order are downloaded, frozen into this repository, and the pages read nothing else. While the site is running there is no outbound request.

Every move order is actually played. The chess rules are implemented here, so each written move is placed on a board in turn. Piece movement, captures, castling, en passant and promotion are all handled, and any move that would leave your own king attacked is discarded. If a single move does not fit the rules the build stops there. Every board you see was therefore genuinely reached.

Families are taken from the names. In the data the name is a family, then a colon, then the variation. So the family is everything before the colon, and before the first comma as well. There is a reason for cutting at the comma. Without it, branches of the same opening ended up separated from their own trunk and looked like distinct families.

There is no page per line. People look for a family, and variations belong inside it. So there is one page per family, and it holds every variation in that family.

The main line of a family is its shortest one, because the short line is where the family branches. If two are equally short the one earlier in the data wins. The rule exists so that no person has to choose what represents a family.

Korean names were added for families only, and even those are ours. Every name in the data is English and there is no official Korean rendering. Variation names stay in English. We would rather leave them untranslated than invent a translation that then looks official.

The boards are drawn on the server. Turn JavaScript off and they are still there, and a search engine sees what a reader sees. The pieces are unicode characters, so not a single image file is involved.

The server counts nothing but how far a visit went. The date, whether the home page opened, whether a family was entered, whether a sibling site was clicked and whether there was a first input, along with the language and a rough source label. Which opening you looked at is not recorded.

Reading the moves

Moves are written in algebraic notation. Squares have names. Files run a to h from the left, ranks run 1 to 8 from the bottom. So one of the squares in front of the white king is e4.

A move is the piece letter plus the destination square. K is king, Q is queen, R is rook, B is bishop and N is knight. The knight is N because the king already took K. A pawn has no letter at all. So e4 means a pawn went to e4 and Nf3 means a knight went to f3.

An x marks a capture. Bxc6 is a bishop taking whatever stood on c6. When a pawn captures, its starting file goes in front, so exd5 is the e file pawn taking on d5.

If two identical pieces could reach the same square, the writing says which. Nbd2 adds the file letter when that separates them, N1d2 adds the rank digit when it does not. If only one piece can get there, nothing is added.

Castling is written O-O and O-O-O. The first is short, towards the king side, the second is long, towards the queen side. A pawn reaching the far rank and becoming another piece is written with an equals sign, as in e8=Q. A move that attacks the enemy king gets a plus sign, and a checkmate gets a hash.

The number and dot in front count the move pairs. 1. e4 c5 means that on move one White played e4 and Black played c5. Each board on this site is captioned with its move number, and a black move is marked with three dots.

Following a family

The boards on a family page follow its main line from the first move. Read those, then read the variation list underneath, and you can see where the same position splits. Each variation shows its ECO code and its full move order.

Large families run to hundreds of variations. The list looks long, but it is ordered by code, so related lines sit next to each other. Reading one code block at a time makes it manageable.

There is a way in from the code side too. If you meet a bare code in a game score, the volume page tells you which families that code covers. That is the place where a code turns back into a name.

Two variations branching at a move means they are the same position up to that point. Learn one family and several of its variations come along with it. That is why reading by family beats memorising lines one at a time.

This site offers no advice on which variation to choose. That judgement involves your opponent, your clock and your taste, and writing it down without a source would be inventing it.

Where people get stuck

If the name you want is missing, the family may be named differently. The same opening carries different names in different countries. Remember the moves instead and come in through the ECO code, which shifts less than names do.

If a Korean name here differs from a book, that is because no standard rendering exists. Our Korean spellings are ours and the English original is always printed alongside. Treat the English name as the reference.

If a board strip stops short of the full move order, it was cut on purpose. One board is sixty four squares, so long lines would make the page heavy. The caption says how far the drawing went.

How often an opening is played is not here. There is no open source for that number. The home page says why.

If you want to know how current the data is, the collection date sits at the foot of every family page. The original repository keeps changing, so ours is the state on that date.