The Hardest Solitaire Games, Ranked by Real Win Rates

Scorpion lets you pick up a face-up card with everything below it, even when that carried group is disordered. The freedom is deceptive. Only same-suit descending runs count toward the finish, there are no foundations to absorb low cards, and empty columns accept Kings alone.

A spectacular group move can relocate disorder instead of reducing it. The solver has proved a floor rather than settled every sampled deal, so the true solvable share is higher than the displayed lower bound. The practical task is to reveal cards without trapping the Kings needed to organize the final four runs.

Canfield exposes only the top card of a thirteen-card reserve. Foundations begin from a random base rank and wrap through King to Ace, while the stock deals three at a time. You are solving the tableau, reserve order, stock packets, and foundation cycle at once.

The solver result is a budgeted lower bound under a capped redeal model, while the playable version allows unlimited redeals. That caveat prevents a clean ceiling claim. What remains true at the table is that every stalled reserve card narrows the position, and an empty column is valuable only when it helps uncover the next one.

Crescent surrounds eight foundations with sixteen fans. Aces build upward, Kings build downward, and both ends wrap. The same rank can look useful on several destinations, so a locally legal move can block the only route that lets the two foundation directions meet.

Three reshuffles rescue some boards, but each preserves the order inside every fan rather than creating a fresh deal. The solver number is a proved lower bound with unresolved searches, not a loss rate. Difficulty comes from coordinating sixteen exposed top cards across two foundation directions before the final shuffle is spent.

Yukon has no stock, so every card is already somewhere in the tableau. You may move any face-up group regardless of its internal order, which creates enormous tactical freedom. The catch is that only the lead card must fit, and the disorder carried beneath it still needs to be repaired later.

Our solver finds many winning lines within budget, so Yukon is not hard because most deals are dead. It is hard because the branching factor is large and attractive moves can bury the one sequence that would have exposed a needed card. Empty columns accept Kings, making early King placement a lasting commitment.

Wasp shares Scorpion’s group-movement freedom but allows any card into an empty column. That one change gives the tableau a workbench and explains why the solver proves far more wins. It does not remove the need to finish four in-suit King-to-Ace runs in place.

The displayed number is a solver lower bound, so it establishes that Wasp is often solvable rather than measuring how often people finish. It remains in the hard set because flexible moves generate many plausible branches. Strong play uses empty columns to separate suits, not merely to move the largest visible pile.

Alaska modifies Yukon by requiring in-suit builds while allowing ranks to run up or down. That sounds flexible, but a sequence can reverse direction only through a carefully chosen join. The board can accumulate long legal groups that point away from the cards they need.

The solver finds a line on many random deals within its budget, so Alaska should not be described as rarely solvable. Its difficulty is cognitive: both directions must be considered while foundations still build upward. Moving a large group is easy; preserving a route to unwind it is the challenge.