Calculation cannot efficiently examine every legal move equally. The practical alternative is to build a set of candidate moves: serious legal moves that deserve concrete investigation or comparison.
A candidate list is a working set, not a sacred quota. It can grow, shrink or change when new information appears.
Legal Moves vs Candidate Moves
Every candidate is legal, but most legal moves are not serious candidates.
A move earns candidate status because it may answer the position's demands. It might:
- force something;
- meet an urgent threat;
- win or save material;
- improve king safety;
- prepare a pawn break;
- simplify favorably;
- improve a poorly placed piece;
- restrict the opponent;
- create a quiet tactical threat.
The purpose of candidate generation is to avoid both first-move fixation and candidate explosion.
Why the First Attractive Move Is Dangerous
The first move you like creates anchoring. Once you start calculating it deeply, the mind tends to search for replies that make it work.
Before investing heavily, ask:
What else deserves comparison?
This does not mean inventing three alternatives mechanically in every position. Some positions genuinely have one legal or practical necessity. The habit simply prevents premature commitment.
A Candidate Funnel
1. Constraint scan
Start with what the position forces you to respect:
- Are you in check?
- Is mate threatened?
- Is a queen or rook hanging?
- Is a pawn about to promote?
- Is there a tactical sequence that must be answered now?
Necessity narrows the search before creativity begins.
2. Forcing-opportunity scan
Look early at moves that constrain replies:
- checks;
- captures with a concrete purpose;
- direct threats;
- tactical transformations such as deflection, clearance, interference or removal of a defender.
These moves often produce narrower trees, which makes them efficient to test. They are not automatically good.
3. Positional and transformational candidates
If no forcing sequence decides the position, broaden the set:
- improve the least effective piece;
- prepare or execute a pawn break;
- improve king safety;
- exchange into a favorable ending;
- maintain useful tension;
- restrict counterplay;
- create a second weakness;
- make a prophylactic move against the opponent's best idea.
Strategy proposes the purpose; calculation tests the concrete move.
Checks as Candidates
Checks deserve attention because they restrict the opponent and can reveal mating or tactical geometry.
But check does not mean good move. A bad check can:
- drive the enemy king to safety;
- lose tempo;
- hang the checking piece;
- force a favorable simplification for the opponent.
Judge the position after the check, not the emotional force of giving it.
Captures as Candidates
Captures change material and geometry. They can remove defenders, open lines or trigger forcing exchanges.
But a capture does not necessarily force a recapture. The opponent may have:
- a stronger check;
- a zwischenzug;
- a counterattack;
- a promotion;
- a tactical threat that makes recapturing irrelevant.
Direct Threats and Tactical Motifs
Recognized forks, pins, skewers, mating nets and defender-removal patterns are valuable candidate generators.
Treat the pattern as a hypothesis. Verify move order, legality and the opponent's best defense before promoting it from "candidate" to "solution".
Defensive Necessities
Candidate generation must include defense when the position demands it.
Defensive candidates may:
- move the threatened piece;
- add a defender;
- exchange the attacker;
- block a line;
- return material;
- create a countercheck or stronger counter-threat;
- simplify into safety.
A defensive move can be the most active move on the board.
Quiet Improving Candidates
Quiet positions often punish players who generate only forcing moves.
Serious quiet candidates can include:
- improving the worst piece;
- centralizing a rook or queen;
- creating luft;
- preparing a break;
- increasing pressure;
- maintaining tension;
- restricting a key enemy piece.
The fact that the opponent has more reply freedom means these lines usually require broader comparison rather than deep forced calculation.
Pawn-Break Candidates
Pawn breaks deserve special attention because they are structurally irreversible and can:
- open files and diagonals;
- create passed pawns;
- fix or dissolve weaknesses;
- alter central space;
- release pieces;
- expose a king.
Calculate the likely recaptures and resulting structure, not just the first pawn push.
Prophylactic Candidates
Ask:
What does the opponent want next?
Then consider whether preventing, discouraging or transforming that idea is worth a move.
Sometimes prophylaxis is direct prevention. Sometimes the best answer is active counterplay that makes the opponent's idea irrelevant.
Candidate Completeness vs Candidate Explosion
A list is incomplete when it omits a move that could plausibly change the decision.
A list has candidate explosion when too many low-value moves remain active, fragmenting attention.
The solution is not a fixed number. Use purpose and urgency to filter.
Serious Candidate vs Pseudo-Candidate
A pseudo-candidate often appears because:
- it gives check but achieves nothing;
- it attacks a piece that moves with tempo;
- it follows a principle while ignoring tactics;
- it looks elegant but fails to the first forcing reply;
- it duplicates another candidate without creating a meaningfully different endpoint.
Many pseudo-candidates can be rejected with a short legality, tactical or purpose test.
Three-Pass Candidate Practice
Pass 1 — no deep calculation
List serious candidates and one reason for each.
Pass 2 — opponent test
For each candidate, identify the opponent's most challenging reply.
Pass 3 — prioritization
Rank which candidates deserve deep calculation first.
Only then calculate the branches.
This isolates candidate generation from later errors. If the best move never entered your list, the tree was doomed before visualization became difficult.
FAQ
How many candidate moves should I find?
As many as the position seriously requires. Forced positions may have one; quiet positions may have several.
Are checks, captures and threats enough?
No. They are a useful search-order family, but serious candidates can also be defensive, quiet, prophylactic or transformational.
Should I calculate every legal move?
No. Candidate generation exists precisely to avoid treating all legal moves as equally important.
A good candidate list is broad enough to contain the right move and narrow enough to let you investigate it seriously.