How to Learn Blindfold Chess for Beginners

A grandmaster playing multiple games simultaneously while facing away from the board... It seems like magic, an innate gift reserved only for the elite. But the truth is far more encouraging: blindfold chess is a highly trainable skill built on structured practice and a solid understanding of the chessboard.

Start Small: Visualizing Micro-Positions

The biggest mistake players make when trying to learn blindfold chess is attempting to visualize the entire board position. This quickly becomes overwhelming. Instead, you must start small by learning to recognize square locations quickly and visualizing moves between those squares.

When you play Quadrant Race, you not only test your square knowledge, but you also train your brain to think of the board in quadrants and subquadrants. After a few days or weeks, when hearing a square name you instantly know its position within its 4×4 quadrant and 2×2 subquadrant and know its colour (because in each 2×2 subquadrant, the top-left and bottom-right squares are always white). Now you can use this understanding of the board in further exercises. (In this chapter, we assume White's perspective for all exercises.)

For example, visualize a knight on e4 — you don't need to try to "see" it in your mind, just know that e4 is the top-left square of the bottom-right 4×4 quadrant and "know" that on that square there is a knight. You don't need to think about a horse shape. Think about a knight in terms of how it moves, what it "feels" like in your mind. Then mentally associate that with the square. Now, name all the squares the knight can move to.

Interlude I: How to Visualize Knight Moves?

Level 0: Text

If you've never tried this exercise, you might calculate the destinations like this:
Because e + 1 = f and 4 + 2 = 6 and 4 - 2 = 2, I know f6 and f2.
Because e + 2 = g and 4 + 1 = 5 and 4 - 1 = 3, I know g5 and g3.

Level 1: Rows and Columns

The previous method can be a useful starting point for beginners, but there is a better way to train your brain: try to use spatial reasoning instead. If you try this for the first time, you may "reason in rows and columns" the following way:

Imagine the right half of the board as four columns. The square e4 is on the leftmost column of those four. (The correct chess term is actually file instead of column.)
If you move two columns to the right in your mind and focus on the third column, do you immediately recognize it as the g-file, or do you still need to consciously translate its position into a letter? If you need conscious translating, don't worry: repeating this exercise for a few days and getting a bit of sleep will change that!
Now imagine the top-right 4×4 quadrant as a big square with four rows inside. Do the same for the bottom-right 4×4 quadrant. (The correct chess term is actually rank instead of row.)
You then "know" or "see" that one of the target squares on the g-file is in the top-right 4×4 quadrant, at the bottom, which is the fifth rank of the whole chessboard. Now translate that back to the name "g5".
Although the calculation 4 - 1 = 3 is simple, don't use it: think about ranks 1 to 4 of the bottom-right quadrant to "see/know" that g3 is the other square on the g-file.

Level 2: Partial Grids

After some practice, you can imagine a grid of 4×4 and "know" the move of a knight within that imagined frame without conscious effort. What you then do:

Translate a square name to a position in an imagined grid (you don't need to "see" it in your mind, but "know" it)
Reason in the imagined grid
Translate back from the imagined grid to a square name

You can identify the other four squares d2, c3, c5 and d6 quite easily.

Level 3: The Full Grid

There are chess players who can imagine the 64 squares at once in their mind. Those players may be able to identify all eight destinations very quickly. For most blindfold chess players, that is actually not the case!

Next, add a target that requires more than one move. Think of a knight on e5. How many moves does it take for the knight to reach d5? Can you find multiple shortest paths? This trains your mind to see the geometry of the board and how different knight routes can be symmetrical in relation to each other. Knight Race was developed for this purpose: playfully do this visualization training and receive immediate feedback on each move.

Let's say you now want to focus on diagonal moves. Maybe you can quickly reason about a bishop on one of the longest diagonals, but can you quickly determine in your mind whether a bishop can move from f2 to b6? You may find it hard to visualize the sequence of squares connecting the starting square to the target square.

Interlude II: How to Reason about Diagonal Moves?

At first, you may need to work through the diagonal step by step, relying partly on text:

Reasoning square by square towards the target square: imagine f2 in the bottom-right quadrant, then e3, then d4 (as the top-right square of the bottom-left 4×4 quadrant), then c5 and finally b6.
Reasoning via the outer squares of the diagonal: "f2 is on g1-a7 and, yes, b6 too".

We propose another system that focuses on spatial reasoning: don't recall each individual diagonal's name as you think, but focus on their lengths and paths in the four quadrants. We propose the following system:

Call diagonals up to length 4 by their length:

Each diagonal has a length. For example, diagonal c1-a3 has length 3. You may remember that as a "length 3 diagonal". All length 3 diagonals are: c1-a3, f1-h3, a6-c8 and h6-f8.
The diagonals d1-a4, e1-h4, a5-d8 and h5-e8 can all be remembered as "length 4 diagonals". Those diagonals go from corner to corner within one quadrant.

Think differently about diagonals that span multiple quadrants:

We think of each diagonal of length 5 as a "131 diagonal" (pronounced "one-three-one diagonal") because the middle 3 squares of the diagonal are in one quadrant and at each end there's one more square in another quadrant. For example: d1-h5 has only d1 in the bottom-left quadrant and h5 in the top-right quadrant, while the middle 3 squares (e2, f3 and g4) are in the bottom-right quadrant. So although the total length is indeed 1 + 3 + 1 = 5, it may stimulate your spatial thinking to remember not length 5, but 1 + 3 + 1, or just "131". When you're in a game, the specific name of the diagonal (e.g. "d1-h5") doesn't matter when you're thinking spatially: you need to understand its structure, you need to visualize it, and then "131" might just help a bit more. For example: when you think of the 131 diagonal that d1 is on, your brain can deduce in an instant that g4 is on that diagonal too. First you think of the position of that square and the name g4 comes automatically (if you've played Quadrant Race enough times).
A length 6 diagonal is also a "222 diagonal" (pronounced "two-two-two diagonal"). For example f1-a6 has two squares in the bottom-right quadrant (f1 and e2), two squares in the bottom-left quadrant (d3 and c4) and two squares in the top-left quadrant (b5 and a6). You don't need to think about each square name, but know this "222" structure instead.
A length 7 diagonal is also a "313 diagonal" (pronounced "three-one-three diagonal"). An example of a 313 diagonal is g1-a7. (Are f2 and b6 on that 313 diagonal?)
Calling the longest diagonals something else is not very helpful, because a1-h8 and h1-a8 are not hard to remember. Perhaps using "the long black diagonal" and "the long white diagonal" is sufficient. You already know they have a 44 structure ("four-four structure").

The more you practise, the more you just "know" or the more quickly you reason. For example, you'll notice that if a 313 diagonal is perpendicular to a 222 diagonal, they always share a square. A pair of perpendicular 222 diagonals never share a square. As your spatial understanding improves, you may be able to "see" any diagonal move (or multi-move path) in your mind at once.

If you are comfortable playing in your mind with single moves on the diagonals, then you may want to do 2-move exercises. Instead of creating your own exercises, you can play Diagonal Race. Be warned: two of the exercises are impossible! (If the target square is a different colour, you must swipe from the centre to the right.)

Ready for an Endgame?

Practising endgames with very few pieces, such as king and queen versus king, is an excellent way to practise doing more moves blindfolded.

Basic Checkmates and Tactics

Once you are comfortable visualizing a few pieces and their moves, you can start doing blindfold puzzles involving easy tactics or mate in one or mate in two.

What if there was a mini-game for this? There is: Puzzle Ascent is designed for training blindfold checkmates and tactics (with fewer pieces).

Another training option of course is to look at a puzzle that is hard, make sure you remember the whole position and then solve it without looking. Maybe it takes half an hour to solve it, but the thrill is probably worth it :)

Full Games, Half Played

When you attempt your first game from the standard position (all pieces), you can focus on the few pieces that are relevant for your next attack or defence action. There is no need to think of all pieces or all squares at once. Between moves, make sure you still remember all piece positions, but don't care too much about visualizing them all in relationship to each other.

Don't expect to finish the game. Instead, set yourself the goal of not finishing the game :)

For example, your goal might be to castle and take both your opponent's knights. Set your own, small goals instead of aiming for a checkmate.

Especially when playing against bots, you can try to find out their weaknesses in a way you would normally not do when playing against humans. For example: do they let a knight be taken if I just move my pawn forward? Do they recognize a fork? You can try out these things while working towards your self-set goal. It doesn't matter if you lose pieces along the way. However, try to remember the positions of all the pieces after reaching your goal. Allow capturing, but not forgetting.

At a later stage, you can set extra hard goals, like capturing ALL your opponent's pieces to force yourself to remember everything. There are always goals at your own level that you can set.

Allow capturing, but not forgetting.

How to Practise Effectively

You don't need to master every exercise before moving on to the next one. The goal is to gradually develop the different skills that make blindfold chess possible.

Here are three principles to keep in mind:

Aim for accuracy before speed. Take your time to reason about the squares and piece positions. As the relationships become more familiar, you will often find that you can work things out more quickly.
Make the exercises challenging, but manageable. If you struggle to keep track of a knight's moves, return to single-square exercises. If you can comfortably solve simple puzzles, try more complex positions. Choose a level that stretches your abilities without overwhelming you.
Don't worry about setbacks. You will sometimes lose track of a position or overlook a possible capture. In the next chapters, we will focus on how to benefit most from your own blindfold mistakes. For now: keep your goals small and return to earlier exercises whenever you need to.

Remember, progress will be gradual. You may decide to take a break for a couple of weeks. When you return to training, you might find that the same exercise has become much easier! Sometimes the brain needs a little bit of time to process your training efforts. By breaking down your training and goals into manageable chunks, you'll gradually develop a wider range of blindfold chess skills. You'll find that playing blindfold chess is a skill that you can build—training by training, move by move.