Five-Letter Word Candidate Filter

Word filter lab

Narrow your candidate list

Paste your words, copy the tile colors, and keep only matching candidates.

Private listDuplicate-aware

Filter result

Remaining words

Formula 1.0.0

3matching candidates

List retained50%

Candidate listFiltered

3 remain3 eliminated

Unique words: 6
Observed rows: 1

Matches from your list

APPLE
AMPLE
ANGLE

3 user-supplied candidates reproduce every entered feedback pattern

Duplicate-aware clue filtering

Five-letter tile feedback is a multiset constraint, so repeated letters must be consumed once

Paste a private candidate list and reproduce the tile pattern from each observed guess. The filter keeps only words that would generate every row under a green-first, remaining-letter-second procedure; it does not load today's answer or claim that a retained word is accepted by the live game.

Step-by-step guide · Duplicate-aware clue filtering

Every clue row is a condition joined by logical AND

Candidate source: Only the 1–500 words you paste
Observed evidence: One to six G/Y/X rows
Duplicate rule: Greens consume copies before yellows

Normalize the private list
Entries are reduced to five English letters and duplicate spellings are counted once. Invalid-length or nonalphabetic entries must be corrected rather than silently reshaped.
Pair guesses and patterns
Each five-letter guess needs exactly one five-symbol G/Y/X pattern. Keep the rows in the same order shown by the puzzle so the audit remains easy to compare.
Recompute feedback for each possible target
The filter runs the same two-stage accounting independently for every candidate and row. It does not merely forbid gray letters globally, because a gray duplicate can coexist with a green or yellow copy.
Inspect the retained set
Zero survivors usually means the supplied list is incomplete or a pattern was transcribed incorrectly. Several survivors mean the observations do not distinguish those words; the result is not a recommendation to choose the first one.

Five-letter tile feedback is a multiset constraint, so repeated letters must be consumed once: A candidate survives only when its generated feedback matches all entered observations in board order.

Resolve exact positions before counting movable letters

Begin a row by marking every green: a guessed letter already in the same position as the possible target. Remove those matched target copies from further consideration. Only then count the unmatched target letters and scan the remaining guessed positions. A letter becomes yellow while an unmatched copy remains; after that supply is exhausted, another occurrence is gray even though the target contains that letter elsewhere.

For example, compare the guess ALLEE with the possible target APPLE. The first A and final E are green. Among the unmatched target letters, only one L remains, so the first unmatched L becomes yellow and the second becomes gray; the earlier unmatched E is also gray because the target's only E was already consumed by its green position. The resulting pattern is GYXXG, not a simple contains-letter test.

Every clue row is a condition joined by logical AND

A candidate survives only when its generated feedback matches all entered observations in board order.

Normalize the private list
Entries are reduced to five English letters and duplicate spellings are counted once. Invalid-length or nonalphabetic entries must be corrected rather than silently reshaped.
Pair guesses and patterns
Each five-letter guess needs exactly one five-symbol G/Y/X pattern. Keep the rows in the same order shown by the puzzle so the audit remains easy to compare.
Recompute feedback for each possible target
The filter runs the same two-stage accounting independently for every candidate and row. It does not merely forbid gray letters globally, because a gray duplicate can coexist with a green or yellow copy.
Inspect the retained set
Zero survivors usually means the supplied list is incomplete or a pattern was transcribed incorrectly. Several survivors mean the observations do not distinguish those words; the result is not a recommendation to choose the first one.

A consistent candidate is not necessarily an official answer

What consistency proves

If a word remains, it reproduces the entered tile feedback under the declared duplicate-aware rule. This is a deterministic claim about the supplied evidence and list.

What the filter never checks

It has no New York Times answer archive, accepted-guess list, account state, date lookup, screenshot reader, language variant, or daily puzzle connection. A typed word can be mechanically consistent but unavailable in the official interface.

Why ranking is separate

Frequency, entropy, answer history, and next-guess strategy require additional data and assumptions. Remaining-count reduction alone does not establish the best next guess or the probability that each survivor is today's answer.

Use neighbouring tools only for the question they actually answer

If you intentionally assign equal prior likelihood to a known finite list and want a simple event fraction, use the Probability Calculator. Do not assume the official answer process makes your private survivors equally likely.
If the task is a FIDE expected-score and K-factor audit rather than word elimination, switch to the Chess Rating Calculator. A rating probability table cannot select a Wordle answer.
For a transparent product of game-specific damage factors, the Battle Damage-Factor Worksheet keeps a numeric modifier ledger. It does not share Wordle's letter-consumption logic.

Keep the puzzle private, current, and unaffiliated

The New York Times controls the current official game, word acceptance, daily solution, branding, and service rules. This independent helper neither retrieves nor reveals that material and is not affiliated with or endorsed by The New York Times. Use only words and feedback you are entitled to enter, and recheck the live game when its behavior or interface changes.

Important limitations

Five-Letter Word Candidate Filter uses only the stated values, rules, or dated dataset supported by this calculator's method. The result is a planning comparison and does not guarantee a price, trade, score, or outcome.

This formula passed its registered engineering tests and is published under an explicit site-owner policy authorization. No independent expert, QA, editorial, legal, or third-party review is claimed. Verify consequential results with a qualified professional.

Duplicate-aware clue filtering

Five-letter tile feedback is a multiset constraint, so repeated letters must be consumed once

Step-by-step guide · Duplicate-aware clue filtering

Every clue row is a condition joined by logical AND

Candidate source: Only the 1–500 words you paste
Observed evidence: One to six G/Y/X rows
Duplicate rule: Greens consume copies before yellows

Normalize the private list
Entries are reduced to five English letters and duplicate spellings are counted once. Invalid-length or nonalphabetic entries must be corrected rather than silently reshaped.
Pair guesses and patterns
Each five-letter guess needs exactly one five-symbol G/Y/X pattern. Keep the rows in the same order shown by the puzzle so the audit remains easy to compare.
Recompute feedback for each possible target
The filter runs the same two-stage accounting independently for every candidate and row. It does not merely forbid gray letters globally, because a gray duplicate can coexist with a green or yellow copy.
Inspect the retained set
Zero survivors usually means the supplied list is incomplete or a pattern was transcribed incorrectly. Several survivors mean the observations do not distinguish those words; the result is not a recommendation to choose the first one.

Resolve exact positions before counting movable letters

Every clue row is a condition joined by logical AND

A candidate survives only when its generated feedback matches all entered observations in board order.

Normalize the private list
Entries are reduced to five English letters and duplicate spellings are counted once. Invalid-length or nonalphabetic entries must be corrected rather than silently reshaped.
Pair guesses and patterns
Each five-letter guess needs exactly one five-symbol G/Y/X pattern. Keep the rows in the same order shown by the puzzle so the audit remains easy to compare.
Recompute feedback for each possible target
The filter runs the same two-stage accounting independently for every candidate and row. It does not merely forbid gray letters globally, because a gray duplicate can coexist with a green or yellow copy.
Inspect the retained set
Zero survivors usually means the supplied list is incomplete or a pattern was transcribed incorrectly. Several survivors mean the observations do not distinguish those words; the result is not a recommendation to choose the first one.

A consistent candidate is not necessarily an official answer

What consistency proves

If a word remains, it reproduces the entered tile feedback under the declared duplicate-aware rule. This is a deterministic claim about the supplied evidence and list.

What the filter never checks

Why ranking is separate

Use neighbouring tools only for the question they actually answer

If you intentionally assign equal prior likelihood to a known finite list and want a simple event fraction, use the Probability Calculator. Do not assume the official answer process makes your private survivors equally likely.
If the task is a FIDE expected-score and K-factor audit rather than word elimination, switch to the Chess Rating Calculator. A rating probability table cannot select a Wordle answer.
For a transparent product of game-specific damage factors, the Battle Damage-Factor Worksheet keeps a numeric modifier ledger. It does not share Wordle's letter-consumption logic.

Five-Letter Word Candidate Filter

Narrow your candidate list

Enter your puzzle state

Remaining words

Five-letter tile feedback is a multiset constraint, so repeated letters must be consumed once

Every clue row is a condition joined by logical AND

Normalize the private list

Pair guesses and patterns

Recompute feedback for each possible target

Inspect the retained set

Resolve exact positions before counting movable letters

Every clue row is a condition joined by logical AND

A consistent candidate is not necessarily an official answer

What consistency proves

What the filter never checks

Why ranking is separate

Use neighbouring tools only for the question they actually answer

Keep the puzzle private, current, and unaffiliated

Important limitations

Narrow your candidate list

Enter your puzzle state

Remaining words

Five-letter tile feedback is a multiset constraint, so repeated letters must be consumed once

Every clue row is a condition joined by logical AND

Normalize the private list

Pair guesses and patterns

Recompute feedback for each possible target

Inspect the retained set

Resolve exact positions before counting movable letters

Every clue row is a condition joined by logical AND

A consistent candidate is not necessarily an official answer

What consistency proves

What the filter never checks

Why ranking is separate

Use neighbouring tools only for the question they actually answer

Keep the puzzle private, current, and unaffiliated