| Rank | Player | W-L | Win % | Games | High Score | Avg Win |
|---|---|---|---|---|---|---|
| 1 | Tom | 391-28 | 93.3% | 419 | 503 (3 bags) | 309 |
| 2 | Luke | 223-21 | 91.4% | 244 | 496 (6 bags) | 308 |
| 3 | Andy | 44-32 | 57.9% | 76 | 435 (5 bags) | 319 |
| 4 | Jon | 41-4 | 91.1% | 45 | 342 | 298 |
| 5 | Other | 13-3 | 81.3% | 16 | 470 | 310 |
A streak resets the day it breaks. These do not.
| Player | Best run | Games | Hands | Tricks | Points | Playing since |
|---|---|---|---|---|---|---|
| Tom | 62 wins Sep 2026 | 419 | 3,106 | 14,467 | 125,883 | Sep 2025 363 days |
| Luke | 50 wins Jan 2026 | 244 | 1,690 | 7,929 | 70,962 | Sep 2025 365 days |
| Andy | 9 wins Sep 2025 | 76 | 491 | 2,762 | 16,058 | Sep 2025 365 days |
| Jon | 23 wins Sep 2025 | 45 | 340 | 1,516 | 13,277 | Sep 2025 17 days |
Three stats that try to be honest rather than flattering. Every one carries the number of games or hands it rests on, because a hot split over ten games is not a fact.
Your wins minus the wins easy Otto would expect against the same mix of Marta levels. Otto is replacement level: the house algorithm playing itself, measured at every rung.
Par is what the hand was actually worth against perfect play, solved from the cards both seats were dealt. Bidding your par is skill; making a bid on a hand worth three more is luck. Solved for 3,204 hands.
Bidding accuracy once someone is within 100 of winning, against every other hand. A swing near zero means the pressure changes nothing, which is what the numbers say.
How each player actually plays, with Otto and Marta from the practice table as the control: that pair is the house algorithm playing itself, so the gap to a human is the human.
| Player | Hands | Exact Bids | Fav Bid | Bags / Hand | Tricks / Hand | Nil | Blind |
|---|---|---|---|---|---|---|---|
| Tom | 3224 | 40.1% | 4 | 0.11 | 4.39 | 23/35 | 54/80 |
| Luke | 1664 | 30.7% | 4 | 0.21 | 4.50 | 63/112 | 87/117 |
| Andy | 494 | 28.3% | 4 | 0.13 | 4.87 | 17/41 | 43/81 |
| Jon | 347 | 34.9% | 3 | 0.20 | 4.37 | - | - |
| Otto (bot) | 4493 | 18.5% | 4 | 1.46 | 4.97 | 1/5 | 248/416 |
| Marta (bot) | 4493 | 19.6% | 4 | 1.44 | 5.03 | 5/12 | 245/395 |
How each regular actually plays, worked out from their own bids, bags and gambles. Titles are earned from the numbers, not assigned. Otto's mirror match is the yardstick: 19% exact bids, 1.5 bags a hand.
Tom bids what the hand will take: exact on 39.9% of hands, twice my own rate, and only 0.73 bags a hand. Nothing is wasted. The price is getting set 17.2% of the time, because there is no cushion under a bid like that.
How to beat Tom: You won't out-bid them, so make the bid cost something: once they've bid, hand them tricks they don't want. Lead high when they're made and let the bags pile up. Their nil is real, so break it early with a low lead in their long suit.
Luke has bid nil 112 times and made it 56% of the time, and goes blind on 7% of hands with a 75% strike rate. A third of the edge comes from the two big swings; the ordinary hands are bid a shade under.
How to beat Luke: Assume the nil is coming whenever they're behind, and duck under everything: lead your lowest cards, follow under theirs, force them to win one. When they go blind, bid normally and set them; a failed blind is a hundred points the other way.
Andy bids a trick more than the hand holds: average bid 4.22, set 27.2% of the time, and the two bids add past 10 on 11.8% of hands. The hand reads fine; the rounding is up, every time. Blind bids are the equalizer: 43 of 81 made.
How to beat Andy: Let the bid do the work. Bid your hand honestly, then play to set them: hold the high cards in their long suits, trump their winners, and don't chase tricks you don't need. When they go blind, that's the hand to be careful in; it's the one they win.
Jon has never bid nil or gone blind in 347 hands, and is set only 12.1% of the time. Safe bids, no swings, no surprises.
How to beat Jon: Bid to your hand and take the tricks they leave behind. They won't punish an aggressive bid, and they won't gamble their way back when behind.
You get a loss only when you finish a complete game and Marta beats you. That's it.
You lose when:
What DOESN'T count as a loss:
Tap the gear next to Marta's name on the game board to pick her level. Every level above easy beat the one below it on the practice table. 14.4% of hands so far were played above easy.
Otto is a bot that plays Marta a few games a day using her own brain, so this is the house algorithm against itself. Bot games live in their own ledger and never touch the leaderboard.
Otto picks a fresh number of games each day, 1 to 100, and Marta ratchets against him the same way she does against a person: up when he wins, down when he loses. Her strength against him settles wherever he wins half the time.