How It Works
Proof of Calculation
ChessRiddle is built on one core principle: you must prove you calculated the full line, not just guess the first move. Most tactics apps only ask for the first move — you can solve them by pattern recognition alone. Our variation tree forces you to build complete lines and consider all opponent responses, training the deep calculation you need in real games. The board stays static while you work, mirroring how you must visualize during tournament play.
Variation Tree
The variation tree is your workspace for proving your calculations. Unlike other tactics trainers that only ask for the first move, you build complete variation trees showing all critical lines — just like annotating a chess book. This forces you to calculate deeply, not just recognize patterns.
The variation tree trains the same mental process you use in tournament games: calculating multiple lines, comparing alternatives, and proving why one move is better than another. This is what separates pattern recognition from real chess strength.
How to Use It
- • Building variations: Enter moves by clicking squares or using the move input panel
- • Navigating: Use arrow keys (← back, → forward) or click any move
- • Branching: When multiple variations exist, right arrow shows branch selector
- • Deleting moves: Right-click a move (desktop) or long-press (mobile) to open the context menu, or press the Delete key
- • Visual feedback: Correct moves highlighted green, incorrect red (after submit)
Keyboard Shortcuts
←Go back one move→Go forward (or show branches)EnterSelect variationEscapeClose branch popupTips
Puzzle Modes
Tactics
Traditional tactics training. Find the forcing sequence — checkmate, winning material, or gaining a decisive advantage. Build your full variation tree to prove you calculated all critical lines.
Why This Matters
Tactics are the foundation of chess improvement. Studies show tactical training is the fastest way to improve from beginner to intermediate level (1000-1800 rating). The variation tree forces you to calculate all lines, not just guess the first move — this is what strong players do.
Tips
Mixed
In Mixed mode, you first classify the puzzle as Tactical or Positional. If your classification is wrong, it counts as incorrect. If it's correct, you proceed to solve normally. Tactical puzzles feature forcing sequences — checks, captures, and threats that win material or deliver checkmate. You must build the full variation tree. Positional puzzles involve quiet improving moves — better piece placement, pawn structure improvements, or prophylaxis. There is no forcing line, just 1 best move.
Why This Matters
Real games blend tactics and quiet moves. Mixed mode trains the crucial skill of position assessment: knowing when to calculate forcing lines vs when to improve your position. This mirrors actual tournament play where you never know what's coming.
Tips
Visualization Modes
Pre-Moves
A sequence of moves is announced — but not shown on the board. You must visualize each move mentally to track where pieces end up. Then solve the tactic from the position you calculated. Trains pure board visualization.
Why This Matters
Calculating moves without seeing them on the board is essential for planning ahead. This simulates the mental process during real games where you must track pieces after your planned sequence. Strong players can visualize 5-10 moves ahead.
Tips
Pieceless
The ultimate visualization challenge. Pieces are invisible on the board. You must reconstruct the position mentally from the piece list, then find the winning tactic.
Why This Matters
Visualization is what separates strong players from masters. Top players can 'see' the board without looking at it. This mode directly trains that skill by forcing you to reconstruct positions mentally — crucial for deep calculation and blindfold chess.
Tips
Opening Blunder
Your opponent just blundered in the opening. Find the refutation! These positions come from real games. The challenge: you don't know exactly which move was the blunder — you need to find it first and enter the punishment.
Why This Matters
Real improvement comes from punishing mistakes, not just solving perfect positions. This mode trains the practical skill of finding refutations in real game positions — exactly what you need in tournaments when opponents blunder.
Tips
Riddle Streak
No hints, no solutions — just you and the puzzles. One mistake ends your streak. Puzzles become more difficult as your streak grows. How many can you solve in a row? Compete on the leaderboard.
Why This Matters
Performance under pressure is a separate skill from solving puzzles. Streak mode trains consistency and decision-making when mistakes have consequences — just like in rated games where one blunder can cost the game.
Tips
Evaluation
You're shown a chess position and must estimate who's better and by how much. Use the evaluation bar or type a precise value. Your guess is compared to Stockfish's evaluation — your guess can be off by up to 30% of the actual evaluation toward equal, and 35% toward the winning side (minimum ±0.5 pawns). For example, if the engine says +3.0, guesses between roughly +2.1 and +4.1 are accepted. Your performance affects your Evaluation Rating — a dedicated rating just for evaluation puzzles. Puzzles are matched to your skill level, so the challenge grows as you improve.
Why This Matters
Position evaluation is a fundamental chess skill. Masters instinctively know when a position favors one side. This mode trains that intuition by forcing you to quantify your assessment — turning a vague feeling into a precise judgment that you can measure and improve over time.
Tips
Practice Mistakes
Practice Mistakes feeds you positions from puzzles you previously solved incorrectly. Instead of moving on, you revisit your weak spots and train until you get them right. The puzzles are unrated so you can focus purely on learning.
Why This Matters
Research shows that targeted repetition of errors is one of the most effective learning techniques. Reviewing mistakes forces your brain to overwrite bad patterns with correct ones — this is how lasting improvement happens.
Tips
Training Plan
Structured practice sessions mixing multiple puzzle types for balanced improvement
What is a Training Plan?
A training plan is a guided session that automatically sequences different puzzle types — tactics, evaluation, visualization — into a structured warmup, main, and cooldown format. You set the duration, focus, and difficulty; the system generates an optimal puzzle sequence.
How to Configure
Navigate to Training Plan from the Training menu. Choose your session duration (30-90 minutes), focus area (Tactics or Visualization), and difficulty level (Easy, Normal, or Difficult). Your last-used settings are saved automatically.
Session Phases
- • Warmup: lighter puzzles to get your mind engaged
- • Main: your chosen focus area at full intensity
- • Cooldown: closing exercises including practice mistakes review
After Your Session
After each session, you can provide optional feedback on difficulty, variety, and length to help calibrate future sessions.
Solution Validation
ChessRiddle uses Stockfish, the world's strongest chess engine, to evaluate every move you play. Your move is compared against the engine's best move — here's how the system decides if your move is correct.
Your Moves
Your move is compared to the engine's best move. If the evaluation difference is within the allowed tolerance, your move is accepted. This tolerance increases with depth: 1.0 pawn for your first move, 1.5 pawns for the second, and 2.0 pawns from the third onward — so your opening move must be the most precise, while deeper continuations allow more flexibility.
"Still Winning" Rule
Even if your move loses more than the allowed tolerance, it's still accepted if the resulting position is clearly winning — specifically, an evaluation of +4.0 or higher. This means you won't be penalized for choosing a different winning path, as long as you maintain a decisive advantage.
Checkmate Moves
Any move that delivers a forced checkmate is automatically accepted, regardless of the mate distance. If you find mate in 5 instead of the engine's mate in 3, both are correct. Opponent responses in mate sequences are also auto-accepted.
Positional Puzzles
Positional puzzles use a tighter tolerance of 0.5 pawns (50 centipawns). Since there's no forcing tactic, the difference between a good move and the best move is more subtle — so the margin for error is smaller.
Opponent's Moves
When you add an opponent response, the system checks how close it is to the engine's best defense. The first opponent move must be within 1.5 pawns, the second within 2.0, and later moves within 2.5 pawns. When you already have a moderate advantage (≥ 3.0 pawns), these tolerances widen further (to 2.5 / 4.0 / 5.0 pawns). With a clear advantage (≥ 5.0 pawns), any opponent defense is accepted. At least one branch must follow the opponent's best defense throughout (all moves within tolerances) to reach the required depth. Any opponent move that falls outside tolerance is treated as a weak response — from that point on, the branch becomes non-critical and all further opponent moves in it are accepted freely.
Solutions That Are Too Short
Normally, submitting a solution finalizes the puzzle and opens analysis mode. With Show Solution Length off, there is one exception: if your submitted moves are on the correct path but just too short, the puzzle stays open so you can add more moves and finish the line you were calculating. The first submission still counts for the puzzle's verdict — but you can verify whether you saw the full idea.
Why This Matters
Understanding these thresholds helps you see that evaluation isn't random — it's a consistent system. If your move is marked wrong, it's because the engine found a significantly better option. Use this knowledge to build stronger variations.
Tips
