Duotrigordle strategy: 32 boards, 37 guesses

Duotrigordle is a budgeting puzzle disguised as a vocabulary puzzle. Thirty-two boards and 37 guesses sounds generous until you count the spare guesses: only five, across the entire game. The useful question is not “what is a good word?” but “which guess keeps five spare guesses alive while reducing the boards that are still expensive?”

The examples below are demonstrations using fixed words from this edition’s 2,542-answer list. Their tile patterns were computed with the same scorer the game uses, but they are not a record of a played daily puzzle. Today’s daily words are not used. The daily sequence is independent of the original Duotrigordle.

1. The real budget: five spare guesses

Normal mode gives 32 boards and 37 guesses. The targets are distinct, so one submitted word can equal at most one target: a single guess can solve at most one board. If every guess solved a board, 32 guesses would finish the grid and five would remain. Every guess that solves nothing—a probe—removes one of those five spare guesses.

The constant is the point: with no probes, the spare count stays at five. Probes change the table. If 20 boards are solved but 25 guesses have been used, 12 boards remain and only 12 guesses are left. The spare count is zero: every remaining guess must solve a board. If 24 boards are solved after 27 guesses, eight boards remain with ten guesses left—two spare.

So a pure probe is affordable only while the spare count is positive. With zero spare, a probe guarantees nothing and leaves fewer guesses than boards.

2. Case: group boards by candidates, not by colors

Four fixed demonstration boards, all still unsolved after the same two guesses, SLATE and STORM. Their feedback looks different, but the important number is how many answer-list words remain consistent with it.

STONE and MOIST each have six non-gray tiles across the two rows, yet one leaves three candidates and the other leaves two. CRANE and BAKER have only three non-gray tiles each, but leave 13 and 30 candidates. Counting colored tiles does not tell you how close a board is; counting the words that still fit does.

The practical consequence is to spend guesses where a candidate guess is also a broad probe. MOIST is one of only two candidates on its board, and its letters M, O, I, S and T are scored on every other unsolved board. Guessing MOIST is a solve attempt and a probe at the same time. A word chosen only to test one narrow board may solve nothing and leave the other 31 boards as expensive as before.

This is a demonstration position: a player does not see the targets, and the candidate sets come from matching the visible feedback against the answer list. The counts are exact for these fixed targets, not a claim that every game reaches these groups.

3. Case: four greens is not a solved board

On the STONE board above, guess STONE. Against the three candidates:

Four green tiles and one gray tile means four positions are right and one letter is wrong. The row has not solved the board. If the target is STOKE or STOVE, STONE leaves both of them possible, and the next guess must separate K from V. A player who counts only green tiles may think the board is finished and move on; the gray N is the evidence that it is not.

With a spare guess, a distinguishing probe is affordable. Against STOKE and STOVE, STERN gives GGYBB for both—it does not separate them. A guess containing K or V does. With no spare, guessing one of the two is a coin flip; the budget, not the color count, decides whether you can afford to find out.

4. When a pure probe is worth one of the five

A pure probe is a guess you do not expect to solve a board with. It is worth considering only when it separates candidates on several boards at once. A probe that separates one pair, on one board, spends a fifth of the game’s spare budget for a single solve. A probe that splits three large candidate groups can save more guesses than it costs.

Use this test before a probe:

1. Count the boards where the probe’s letters are still unknown.
2. Count how many candidate groups it would split, not just how many green tiles it might produce.
3. Check the spare count. If it is zero, the probe cannot be afforded; every guess must solve a board.

In the four-board excerpt, the broad groups are CRANE with 13 candidates and BAKER with 30. A guess that separates several of those candidates while also testing letters on the other boards can be worth a spare. A guess aimed only at MOIST versus MOTHS is harder to justify: it is one board, and one of the two words is already a candidate that scores everywhere else.

5. Where Duotrigordle sits

Each edition has its own answer list and daily sequence, so the same date does not mean the same words.

Source and verification note

The demonstration patterns and candidate counts were checked on 29 October 2026 with this edition’s own letter-feedback scoring, its 2,542-word answer list and its 14,855-word accepted-guess list. The examples are demonstrations, not records of a real player’s game, and no claim is made that they are optimal play.

Mode rules · Tile feedback · Play today’s Duotrigordle · Practice