Quick answer
Sort the deal, count which card category the most tiles already serve, find a pair prospect, and let the Charleston choose between your top two directions.
Thirteen tiles land in your rack and the card has dozens of hands on it. Every new player asks the same question: which one? Experienced players do not scan the card line by line. They read the deal for direction first, then check the card. Here is that method, step by step.
Sort the deal so it can tell you something
Group your tiles by suit, then by number inside each suit, with winds, dragons, flowers and jokers off to one side. A rack sorted this way shows its own shape: runs of neighbouring numbers, matching pairs, three of a kind, and the gaps between them.
Do this before you look at the card. A sorted rack asks a better question of the card than an unsorted one, and it takes ten seconds.
Count toward categories, not toward hands
The card is organised in sections, and each section is a family of hands built the same way: consecutive runs, even numbers, odd numbers, quints, winds and dragons, singles and pairs, and so on. Count how many of your tiles could serve each family. A deal with 6-6-7-8 in one suit and a 7 in another already has six tiles pointing at consecutive runs, whatever the exact line turns out to be.
Two or three families will usually score higher than the rest. Those are your candidate directions. Only now is it worth reading the specific hands in those sections of the card.
Find your pair prospect
Most hands on the card need at least one pair, and jokers cannot be used in a pair. So a direction that needs a pair you do not have and cannot reasonably draw is weaker than it looks on the count alone.
If your deal already holds a natural pair that fits one of your candidate families, that family moves up. If your best-counting family needs a pair of a tile you have not seen, note that as its risk.
Let the Charleston decide, not the deal
The Charleston will move nine tiles through your rack before the first discard, so the deal is a starting point, not a verdict. Pass the tiles that serve none of your candidate families. Keep anything that serves two of them.
By the third pass of the first Charleston you should be able to name your top two directions. Use the second Charleston to confirm the leader, or to switch if the incoming tiles clearly favour the backup.
Keep two directions open until the tiles close one
The commonest beginner mistake is committing on the deal to a single hand and then passing tiles that would have made a better one. Hold your top two families through the Charleston and let the draws break the tie.
A useful test: if a tile serves only your second direction, keep it while the two are close and pass it once the leader is clearly ahead. Flexibility is worth more than a fast start.
Commit, then play the table
Once one direction is clearly ahead, commit. Now the tiles from the losing direction are safe to discard, and your attention shifts to what other players expose and discard, so that you do not throw the tile that completes someone else's hand.
This whole sequence, from sorting the deal to naming the leader and the backup, is what the coach does out loud when it reads a hand. Reading a few dozen hands that way is the fastest route to doing it at the table without thinking.
Put it into practice
A fresh coached hand every morning, the coach on call for your own tiles, all ten lessons and the 14-day course. $14.99 a month, cancel anytime.
You have just read the method. Mei, our AI coach, runs it on your actual tiles: which category leads, what the backup is, and what to pass, explained in plain English.