Mancala Game

Six pits per side

Each side begins with six pits and a store. Four stones start in every pit, for forty-eight stones total. On your turn, choose one non-empty pit on your own side, pick up all of its stones, and sow them one at a time around the board. Your own store receives stones, but the opponent’s store is skipped. The player with more stones in their store after the final sweep wins.

This sounds simple, but move order matters immediately. A pit that looks weak can be strong because the final stone lands in your store and gives another turn. A capture can look attractive but open a larger reply. The board therefore rewards counting, tempo, and planning rather than only collecting the largest visible pile.

Extra turns and captures

If your last stone lands in your own store, you play again. Chaining extra turns is one of the central skills in Kalah. If the last stone lands in an empty pit on your own side while the opposite pit contains stones, the landing stone and the opposite stones are captured into your store. The Strategy Coach calls attention to these ideas without pretending that a single heuristic is always correct.

When either player’s six pits are empty, the game ends and all stones remaining on the other side sweep into that player’s store. That sweep is why late endings can reverse an apparently comfortable lead. The exact tactic trainer is built around these sparse endings and solves them all the way to the final store count.

Count the landing point

The first useful calculation is where the last stone will land. If it reaches your store, you keep the turn. If it lands in an empty pit on your side, check the opposite pit for a possible capture. If it travels far enough around the board, remember that the opponent’s store is skipped. A move that begins with many stones can therefore have a landing point that is not obvious at a glance.

Next, look at the board you leave behind. Giving the opponent an easy store landing can hand them two moves in a row. Leaving a valuable pile opposite one of their empty pits may create a capture. Strong play often comes from denying those replies even when your own move gains fewer stones immediately.

Use the endgame trainer to verify intuition

Late positions are especially good for learning because the remaining game tree is finite enough to solve exactly. In the tactic trainer, a legal but suboptimal move does not alter the board. The miss is saved to Failed Review and you can try again from the same position. That separates “I knew the rule” from “I chose the best sequence.”

The displayed exact margin is not a heuristic score. It is the final store difference after both sides play optimally from the original position. The gap tells you how much worse the next-best legal first move would be. That makes the trainer useful for discovering when an extra turn, capture, or final sweep matters more than it first appears.

Study tools are allowed, but labeled

Hint and Undo are learning tools, so using either one marks the current run Assisted. An ordinary assisted win still counts as a played result, but it cannot create a Clean best margin. Undo keeps the Assisted label instead of restoring an older Clean snapshot. This prevents a player from taking a hint, undoing, and then accidentally saving that run as an unaided personal best.

No server account is required

Practice statistics, Failed Review, autosaves, best margins, and Daily specialist history are browser-local. The current public build does not use an account system, global leaderboard, cloud sync, paid API, or backend database. Clearing browser storage can remove these local records, so the site describes them as local records rather than permanent online achievements.