Memory Chess

Memory Chess is a free memory chess game and chess memory trainer that puts a number on your mind's eye. A round works like this: you choose how many pieces appear on the board and how long you get to study them, the position is displayed for that window, the board goes blank, and you rebuild the position from memory. On submit you receive an accuracy reading from 0 to 100% plus a map of every square you got right or wrong. It runs in the browser on desktop and mobile, needs no account, download or sign-up, and is available in 24 languages. Most chess practice trains decisions. As the site puts it, very little of it checks whether you can actually hold the board in your head, even though that skill sits underneath calculation, blindfold play and spotting tactics and is usually left to chance. The observation behind Memory Chess is concrete: three moves deep into a calculation, a bishop on the long diagonal quietly disappears from the picture in a player's head, and the board in front of them was right while the one in their mind was not. The product's hypothesis is that board recall is a trainable, testable skill. Rather than making claims about your brain, it holds a position briefly, has you rebuild it, checks the result, and lets you repeat with more pieces and less time so you have a reading you can track. One round, one clean number. You pick a piece count between 2 and 32 and a study window between 1 and 60 seconds, either through a preset or the custom rig, then press start and study the position until the meter empties. When the board clears there are no pieces and no hints: you select a piece from the palette, tap a square to place it, and tap a placed piece to remove it. Submitting compares your rebuild against the original and returns an accuracy percentage along with a square-by-square map, where correct squares, wrong placements and missed squares are each marked and nothing is hidden. Placing more pieces than the position actually had lowers your accuracy, so the game rewards only placing what you genuinely remember. Because a round takes under a minute, you can run many of them. Difficulty is expressed as two variables, with load going up while exposure goes down. Easy is 2 pieces with 10 seconds, described as just kings and perfect for starting out. Medium is 6 pieces with 10 seconds, a few pieces with comfortable time to memorize. Hard is 12 pieces with 8 seconds. Grandmaster is 20 pieces with 5 seconds, aimed at chess masters with photographic memory. The custom rig exposes both variables directly, 2 to 32 pieces and 1 to 60 seconds, so you can isolate a single weakness; the site's advice is to change one setting at a time and to move up a tier only when your readings hold steady rather than after one lucky round. A separate calibration reading loads the Medium round from a real game, 6 pieces including two kings with 10 seconds of exposure, and produces a readout card with correct squares, rebuild time, wrong pieces, exposure and recall by piece type, plus a suggested difficulty. Calibration works with mouse, touch and keyboard. Every round you play is recorded on the device and feeds a dashboard, because one reading is a snapshot and ten readings are a measurement. The figures cover memory span, pieces held at any setting, accuracy over time round by round, recall pace in seconds per piece, days in a row, a forgetting curve, a lab notebook of what happened and when, a miss map showing where your squares slip, recall by piece type, insights, programs, a goal tracker and personal bests per setting. Panels start as sample values and switch to your own record once you have played, and a panel that cannot read your record yet says what it still needs. Insights only appear once their numbers pass a minimum sample, and they name the counts behind them, for example that the a and h files were missed about 3.5 times as often as the centre files. Miss maps and piece-type breakdowns point to specific weaknesses, such as pawns and knights slipping first, and the pages link drills that address them. Several features turn isolated rounds into a practice routine. Today's board is one position a day, the same for everyone, opening at midnight UTC with a fixed study time and a single try, and the homepage shows how many pieces it has and how long you get. A streak counts any finished round, game or practice, for its day, with one missed day in any seven forgiven. Boards you score under 80% on return for review 1, 3, 7 and 14 days after you first see them, plotted as a forgetting curve that the site labels an illustrative model of the spacing effect rather than measured data. Programs give you one plan at a time: a seven-day Baseline week of one Medium round a day, a fourteen-day Edge-file drill at 8 pieces and 15 seconds reviewed against your miss map every three days, and an open-ended Step-down ladder that holds 90% for three rounds before cutting study time by two seconds and eventually adding pieces. Goal and personal bests track a target piece count and accuracy, and bests are kept per setting because 6 pieces and 20 pieces are different tests. The Learning Center walks through a full round on a real position, the final position of Kasparov's Immortal after 44.Qa7, as a worked example of one way to memorise a board rather than a recording of anyone's attention. It proceeds in phases: study eleven pieces for ten seconds and take in the whole board rather than square by square; fix anchors, starting with the two kings; chunk the eleven pieces into three ideas grouped by what they do to each other; encode each group as a sentence, Stand-off, Stairs and Long shots; let the board clear with nothing to do but hold three sentences; rebuild group by group, placing each group with its sentence before starting the next; then check the rebuilt board against the relations in your sentences before scoring. The example finishes at 91%, ten of eleven squares, and shows how a relation check catches a slip that a piece count alone would miss. The site cites de Groot's experiments, Chase and Simon (1973) and Gobet and Simon (1996) as background, and states plainly that the sentences themselves are a technique, not a research finding. The headline benefit is feedback: you cannot improve what you never measure, and Memory Chess replaces vague impressions of board vision with a number you can repeat and compare. Each round isolates recall from decision-making, so a weak result is not confused with a bad move. Because readings are stored per setting and plotted over time, progress and plateaus become visible, and the miss map and piece-type breakdown turn a general feeling of losing track into a specific file or piece to practise. Short rounds make regular practice practical, and spaced reviews put the boards you failed back in front of you at increasing intervals. Concrete scenarios follow from the settings. A beginner who does not know openings or calculate moves starts on Easy with two kings and ten seconds, then adds pieces or shortens time. A player who wants a baseline takes the calibration reading and keeps the readout card. Someone building a habit plays the daily board and protects a streak. A player whose misses pile up on the h-file runs the custom rig at 8 pieces and 15 seconds and reviews the miss map every three days. Visualization and blindfold training are served by the library guides on board vision, visualization, calculation, memory chunking and a four-stage blindfold plan, each paired with drills. Competitive players submit results to the leaderboard, which ranks by correct pieces, then fewer wrong pieces, then faster memorize time, then faster solve time, and shows each player's country. Memory Chess runs in the browser on desktop and mobile, is free to play, and requires no account. There is no server: your record is kept on the device, and you can download a copy of your lab record to move it to another browser or keep it safe, then import it, while backing up across devices is listed as coming soon. Adding a name to the leaderboard is optional, and players can also swap scores on Telegram. The site reports 48,697 games played. The reference library pairs short guides with drills so every reading points to something you can practise. Memory Chess turns chess memory into a short, measurable drill: study a position, rebuild it, read the number, and come back for the trend line. It is free, needs no sign-up, and gives players a way to test and train the board recall that sits underneath calculation, visualization and blindfold play.