Scrabble Cheating Detector

Analyze a suspicious Scrabble move and get a Suspicious Move Score based on how unusual and solver-like the play was. Set up the position before your opponent's move, mark the tiles they placed, and compare the play to strong solver output and to typical human choices.

  1. Board
  2. Mark move
  3. Score

How do you want to enter the board?

Upload a screenshot to fill the board automatically, or enter it by hand. Either way you'll review the board before anything is analyzed.

How the analysis works

  1. Reconstruct the board as it was before your opponent's move.
  2. Have a game solver evaluate the strong, legal plays from that position.
  3. Compare the observed move against patterns from human gameplay.
  4. Account for the tiles on the opponent's rack that you can't see.
  5. Combine these into a single Suspicious Move Score.

See a fuller walkthrough of the method →

Understanding the Suspicious Move Score

The Suspicious Move Score (SMS) runs from 0 to 100 and summarizes how unusually solver-like a move is compared with human play. Low scores mean the move looks ordinary; high scores mean it closely matches what a solver would choose and is rarely picked by human players.

Scrabble has deep, well-studied strategy around rack leaves and board control. The score considers whether an opponent's choices track solver preferences — including subtle defensive or setup plays that most human players overlook.

A high score on a single move is not proof of cheating — luck and skill both produce strong plays. Repeated high scores across many moves are much more meaningful than any one result.

Frequently asked questions

Can a Suspicious Move Score be used as proof of cheating?

No. It is an analytical signal, not a verdict. Treat a high score as a reason to look closer, ideally across multiple games, rather than as evidence about any single move or player.

Why are repeated solver-like moves more suspicious than one?

Any one strong move can come from skill or luck. But finding the solver's preferred play again and again — including obscure words and non-obvious defensive moves — becomes statistically unlikely for unaided human play as the sample grows.

Does a high rating make a low score more likely?

Stronger players naturally produce more solver-like moves, so context matters. When a rating is available it can be factored in, so an expert's strong play is judged against expert expectations rather than beginner expectations.