Almost every FreeCell deal can be won. Of the 32,000 original numbered deals, exactly one — #11982 — has been proven impossible. So when you lose, you were beaten by a rectangle of cardboard, and that is a useful thing to sit with.
You've seen the game let you move five cards at once, then refuse to move three. It isn't being capricious. FreeCell never really moves a stack at all — it moves cards one at a time, through your free cells, very fast, and then shows you the result. So the number you can shift is exactly the number you could have shuffled through by hand:
(free cells + 1) × 2 for every empty column
Four cells free and no empty columns: five cards. Four cells and one empty column: ten. Four cells and two empty columns: twenty. That doubling is not a small thing — it is the entire game.
Beginners hoard free cells and fill columns. Strong players do the opposite. A free cell adds one to your capacity; an empty column doubles it. If you can trade a cell for a column, take the trade nearly every time.
And an empty column does something a cell can't: it holds a sequence. You can park an ordered run of six cards in a column. You cannot park anything in a cell but a single card, where it sits and sulks and reduces everything you can do.
1 – Get the aces out, but not at any price. An ace buried under four cards is worth digging for. An ace that costs you three free cells to reach is a trap, because a full cell is a dead cell.
2 – Don't send cards home too early. This is the one that surprises people. A two of hearts on the foundation is a two of hearts you can no longer use to hold a black ace. Foundations are a one-way door. Late in the game they're a relief; early, they're a loss of flexibility. If a card is still useful on the tableau, leave it there.
3 – Look for the long run before the easy move. The obvious move is usually available in ten seconds and gains you nothing. Spend a minute looking for the sequence of six moves that empties a column. That's where the game is won.
4 – Count before you commit. Before you start a manoeuvre, count how many cards you'll need to move at the deepest point, and check the capacity formula. Half of all FreeCell losses are someone starting a plan they didn't have the cells to finish.
You're probably not. The deal is almost certainly winnable and the solution is almost certainly a move you dismissed. Two things to try:
Unlimited undo is right there and nobody is watching. Go and lose a few; that's how the pattern-recognition gets built.
Free cells plus one, doubled for each empty column. Four cells and no empty columns lets you move five cards. Four cells and two empty columns lets you move twenty. The game shows you the live number so you don't have to work it out mid-move.
Very nearly. Of the classic 32,000 numbered Microsoft deals, only game #11982 has been proven unsolvable. Across random deals, well over 99% can be won — so when you're stuck, the answer is in there.
No, and this is the mistake that costs most games. A card on the foundation can never come back. Early on, a low card is often more useful on the tableau holding something else. Send cards home when they've stopped being useful, not when they first become eligible.