Every screen, control and feature, one at a time.
You sit South. The robots play the other three seats using the bidding system and conventions you choose in Settings — American Standard 2/1 or ACOL, each with customisable conventions.
Across the top, three chips show the rubber at a glance — games, we total, they total — and tapping any of them opens the full score pad.
Below them:
Deals continue until one side wins the rubber; then the app offers to start the next one. The rubber, your settings, the archive and the live scorepad are all saved on the device as you go, so closing the app loses nothing.
The bidding panel fills the lower half of the screen during the auction. Its header shows your system, and your hand's high-card point count.
The auction grid shows every call so far in four columns, one round per row, with the dealer's column marked. Tap any call — yours or a robot's — to see what it meant. This is the fastest way to learn: the robots explain every bid they make, in your system.
To make a call: pick a level, then a strain. The meaning appears above the buttons before you commit, and you place it with the Place button. If you already know what you want, double-tap the strain (or Pass/Dbl) to bid it straight away.
💡 suggests a call — the engine's own choice for your hand, with its reasoning, pre-selected so you can place it or overrule it. There is no penalty for using it, and no limit.
Illegal calls are disabled rather than hidden, so the shape of the auction stays readable. Changing your bidding system or a convention while an auction is live restarts that auction, because half a hand bid in one system and half in another would be meaningless.
When the auction ends, dummy comes down and the play begins. Your legal cards are bright and tappable; illegal ones are dimmed, so you cannot revoke.
When your side declares, you play both hands — your own and dummy's — even on deals where North is nominally declarer. The table rotates so that whichever of the two hands is to play sits at the bottom. When the robots declare, you defend with South's cards and see dummy as everyone does.
The pills above your hand carry the tools:
Trick counters under the table show WE and THEY. When the thirteenth trick falls, the result sheet appears with the score, any games or rubber won, and a one-line verdict on the deal.
Hint behaves differently in each stage, and the difference matters:
Undo steps back one of your cards, cancelling any robot moves that followed. There is no limit within a deal and no penalty. It is meant to be used: playing a card, seeing what happens and taking it back is a faster way to learn a suit combination than reading about it.
Claim and Concede rest appear only when a double-dummy solve of the remaining position proves the outcome, so accepting one never costs you a trick. Concede is also on the ≡ table menu at any point, for when you simply want to move on.
Nothing you use here affects the score. The Analysis screen grades what you did, not how you arrived at it — though the sensible way to use the app is to bid a hand yourself, then check the hint afterwards.
After every deal, Analysis re-examines the hand with all four hands visible. It takes a few seconds, because it is running a large number of double-dummy solves, and it is the most valuable thing in the app.
The Deal panel at the top shows all four hands, the contract and the result, plus a table of how many tricks each side can take in each strain — the double-dummy trick table. Under it sits the par result: what the deal is worth if both sides bid and defend perfectly. Par is the honest yardstick for the auction, because it accounts for the fact that some deals belong to the opponents.
Your grade follows, with its two halves — bidding and play — and, when you open it, the reasoning behind each number.
What to work on gives a plain-language verdict on the deal and then the specific lessons, strongest first: the call that cost the most, the trick that cost the most.
The Auction, Reviewed lists every call against the engine's recommendation for your system and vulnerability, with the reasoning for each. Only your own calls are marked right or wrong; the robots' calls are shown for context.
The Play, Reviewed lists all thirteen tricks with the winner of each. Cards are annotated against the double-dummy best play, and tricks that actually changed hands are marked in place, naming the card that should have been played. Again, only cards you chose are graded.
Replay This Hand deals the same cards again, unscored, so you can try the line you now know was right.
The percentage on the Analysis screen is not "how many of your decisions matched the engine". It is a measure of what your mistakes cost, in points, and it is built from two questions a bridge teacher would ask.
1. Bidding — was the contract you reached worth what the cards were worth? The app computes par for the deal and compares it with the value of the contract you actually reached, played double-dummy. It has nothing to do with how the cards then fell: reaching the right contract and getting unlucky scores full marks.
2. Play — how many tricks did you give away? The exact double-dummy trick count is tracked at every trick boundary through the whole hand. Each time it moves against your side, a trick has just been lost; the app attributes it to the seat that caused it and charges you only for the ones you were responsible for. Your partner's and the robots' errors are never charged to you.
Both are measured in points, and points become a percentage through the IMP scale — bridge's own compressive measure of how much a result differs from another. A one-trick overtrick difference is 1 IMP; letting a cold vulnerable game go down is around 12. A 16-IMP loss scores zero, and everything below scales linearly, so each IMP costs a little over six points.
| Percentage | Label |
|---|---|
| 95–100 | Expert |
| 85–94 | Strong |
| 70–84 | Solid |
| 50–69 | Developing |
| Below 50 | Keep practising |
The design consequence is deliberate: a hand where you threw away a trick can never read as perfect, no matter how tidy the endgame was. Equally, a hand where nothing was available and you took what there was scores highly even if the result looks bad on the pad.
Every completed deal is saved automatically — including passed-out deals and practice replays — with its contract, result and grade. The most recent 50 are kept; older ones drop off the end.
Tap any row to open its full analysis, exactly as it was after the deal. The analysis is computed fresh, so an archived hand from a rubber three days ago gives the same trick table, par and grading as it did at the time.
Individual hands can be deleted with ✕, or the whole archive cleared. Both are permanent — there is no undo for the archive itself.
The archive is the practical way to use the app as a teacher: play a rubber without stopping to analyse, then work through the deals afterwards and look for the pattern in what the grades are telling you.
For a game with real cards and real people. It scores a rubber correctly so nobody has to argue about the arithmetic, and it is entirely separate from the robot game's score.
To record a hand: choose whether it was ours or theirs, the level, the suit, whether it was doubled, and the result — = or +n if it made, or the number of undertricks if it went down. Add honours if a hand held them. A preview line shows what the entry will score before you commit it.
The pad then keeps everything in step: the below-the-line total, the games, the vulnerability, the rubber bonus when it is won, and the unfinished-rubber premiums if you stop early.
Corrections are safe. Tap ✎ on any recorded hand to edit it, or ✕ to delete it — the entire rubber is re-scored from scratch, so games and vulnerability stay right even when you correct the second hand of six. An edited hand is re-scored at the vulnerability it was played at, not the one the rubber has since reached.
There is a timer for the rubber, a Save Part-Finished Rubber option that applies the 300/100 premiums correctly, and a list of past rubbers with their durations and results. Resetting the live scorepad does not touch the robot game.
Base system — American Standard 2/1 or ACOL. Choosing one resets the convention switches to that system's defaults, and restarts any auction currently in progress.
Convention switches — the full list is in the Bidding chapter's reference table. Every switch changes the robots, the meanings shown when you tap a call, the hint, and the Analysis screen's judgement of your auction, because all four come from the same engine.
High-contrast four-colour suits — red hearts, black spades, blue diamonds and green clubs on the card faces, with a live preview. Useful on a small screen, where black spades and black clubs are easy to confuse in a fanned hand.
Pace of play — Relaxed, Normal or Fast. This sets how long the robots appear to take before each bid and card. It does not make them play better or worse: the thinking runs alongside the pause rather than after it, so the wait is the longer of the two, never their sum.
The app plays fully offline. There is no account, no sign-in and no server: the bidding engine, the card-play engine and the double-dummy solver all run on your device.
Nothing about your play leaves the device. Your rubber, settings, hand archive and live scorepad are stored locally, which also means they belong to this installation — deleting the app deletes them.
Double-dummy analysis is powered by DDS by Bo Haglund and Søren Hein, used under the Apache License 2.0.
Rase plays a full rubber, scores it the traditional way, and grades every bid and card afterwards.
Free, with no ads, no accounts and no tracking. The app plays offline.