A Strategy Guide to Word Circuit

Word Circuit is a five-letter deduction board we wrote ourselves. Six tries, three signals, one answer a day. Because it is our own code, this guide can be specific about how the scoring works rather than guessing at it, and a surprising amount of good play follows directly from that.

The three signals

After each guess, every letter comes back as one of three things. Exact means that letter is in the answer and in that position. Near means the letter is in the answer somewhere else. Out means it is not in the answer at all — with one important qualification, covered below.

The scoring runs in two passes, and the order matters. First it walks the whole guess and marks every position where your letter matches the answer’s letter exactly. Only the answer letters that did not get an exact match go into a pool for the second pass. Then it walks the guess again and, for each unmatched position, checks whether that letter is still available in the pool — if it is, the letter comes back near, and one copy is removed from the pool.

That two-pass structure is why the board occasionally looks like it is lying to you.

The repeated letter trap

Suppose the answer contains one E, and you guess a word with two of them. If one of your E positions is exact, the pool has no E left, so your other E comes back Out. Out, in that situation, does not mean “there is no E in the answer” — it means “there are no more E letters than the ones already accounted for”.

This catches people constantly, and it costs whole guesses when players cross a letter off mentally that is actually sitting right there in an exact square. The rule to carry: a grey letter only tells you something new if you have no exact or near mark on that same letter anywhere else on the board.

It cuts the other way too. If you guess a word with two of a letter and both come back near, the answer contains at least two of them, which is a much stronger piece of information than a single near mark and worth planning around.

Repeated letters are rare in the answer set — only three of the answers contain one — so on balance you should assume five distinct letters until the board tells you otherwise.

The guess list is closed

Word Circuit does not accept any five-letter string. There is a fixed list of allowed guesses, and anything outside it is rejected with a message rather than being scored. Crucially, a rejected guess does not consume a try, so there is no penalty for trying a word and finding out it is not there.

The practical effect is that pure information-gathering strategies are limited. In games with a huge dictionary you can afford to burn a guess on a word full of untested letters that you know cannot be the answer. Here the list is compact enough that most words which would make good probes are also plausible answers, which is a nicer position to be in — your probe can win.

Which opener to use

The answer set is small and hand-picked, and it has a distinct texture: a lot of workshop and electrical vocabulary sitting alongside natural imagery. That skews the letter distribution in a way you can exploit.

Counting across the answer pool, E and R each appear in sixteen of the answers, A in fourteen, L in twelve, and I in eleven. N and O each turn up in ten. Anything without at least three of that group is a weak first guess.

CRANE is the strongest opener we have found. It covers C, R, A, N and E, and at least one of those letters appears in twenty-eight of the thirty possible answers — so an opener that comes back completely grey has already eliminated almost the entire pool, which is an excellent outcome disguised as a bad one. It is also itself a possible answer, so it can simply win.

STONE, TRAIL and EARTH are reasonable alternatives if you want to rotate. All three hit the high-frequency group and spread their letters across different positions.

Guesses two and three

The common mistake at this stage is chasing the letters you have found instead of testing the ones you have not. If the opener returned two near marks, you have three unknown slots and four untested high-value letters. A second guess that repositions your two known letters and introduces three new ones is worth far more than one that only shuffles what you already have.

Position information is worth as much as letter information. Near tells you a letter exists and that it is not where you put it, which removes one of five slots. Two near marks on the same letter across two different guesses narrow it to three slots, then two, and the answer often falls out of position logic before you have identified the last letter at all.

One structural hint that is fair game: the final letter is E in nine of the thirty answers, more than any other ending. If you are down to two candidates on the last try and one of them ends in E, that is the way to lean.

When you are down to the last try

Stop optimising and commit to the most likely word. There is no partial credit, and a clever probe on guess six is a guaranteed loss rather than a possible one. After a loss the board offers to reveal the answer, which is worth taking — the answer set repeats on a cycle, so a word you have seen once is a word you will see again.

If you want a different kind of word board afterwards, Mini Clue Grid is the other one we make, and Word Cross and Word Connect Puzzle sit in the wider puzzle collection.