In most trick-taking games the answer would be "the highest trump". The Fox in the Forest is different: taking ten or more tricks scores zero, and taking three or fewer scores as much as winning. So a card is good when it gives you control, the ability to win a trick or dodge it when you choose. That is exactly what the numbers show.
What a card in your starting hand is worth
Points per round compared with an average card. Trump means the suit of the decree card.
- As trump
- Off-suit
Show the numbers
| Rank | As trump | Off-suit |
|---|---|---|
| 1 | -0.16 | -0.22 |
| 2 | +0.19 | +0.30 |
| 3 | +0.54 | +0.74 |
| 4 | -0.05 | -0.16 |
| 5 | -0.04 | -0.09 |
| 6 | -0.40 | -0.33 |
| 7 | +0.27 | +0.17 |
| 8 | -0.04 | -0.11 |
| 9 | -0.12 | -0.26 |
| 10 | -0.17 | -0.07 |
| 11 | -0.09 | +0.08 |
Why the Fox wins
The Fox lets you pick the trump suit after you have seen how the round is going. Short on tricks? Make your longest suit trump. Trying to stay humble? Make the suit you barely hold trump, so you can't win by accident. It also changes trump for the trick in progress, which can flip a trick you were about to lose or win.
The bots agree with how they play it. They hold the Fox for a while (average trick 6.4 of 13 off-suit) and swap the decree 64% of the time they play it; 53% of plays actually change the trump suit. As a trump it is played earlier (trick 5.1).
Why the 11s don't matter
A high card wins tricks, and winning tricks is only half of this game. Holding several of them pushes you toward ten or more tricks, which scores zero. The Monarch (11) does force your opponent to give up their best card of the suit, but our measurements put it at +0.08 off-suit and -0.09 as trump: a wash.
The Witch (9) is a nice illustration. Off-suit it wins 96% of the tricks it is played in, which is great right up to the moment you don't want the trick. That makes it a below-average card: -0.26.
The worst card: the plain 6
The 6 has no ability and sits in the worst spot: too high to lose a trick safely, too low to win one. It scores -0.33 off-suit and -0.40 as trump. The 2 is the mirror image: no ability either, but low enough to duck almost any trick, and it comes out at +0.30.
Every special card, ranked
When strong bots play each special card
Off-suit plays only, forced plays left out. Early = tricks 1-4, middle = 5-9, late = 10-13.
- Early
- Middle
- Late
Show the numbers
| Card | Value off-suit | Average trick | Led it | Won the trick | Plays |
|---|---|---|---|---|---|
| 1 Swan | -0.22 | 6.0 | 54% | 5% | 6,242 |
| 3 Fox | +0.74 | 6.4 | 54% | 31% | 6,496 |
| 5 Woodcutter | -0.09 | 6.8 | 47% | 38% | 6,064 |
| 7 Treasure | +0.17 | 5.1 | 63% | 59% | 6,816 |
| 9 Witch | -0.26 | 6.2 | 50% | 96% | 5,830 |
| 11 Monarch | +0.08 | 6.8 | 49% | 65% | 5,994 |
- 3 Fox (+0.74 off-suit, +0.54 as trump). Swap the decree card with a card from your hand, changing trump on the spot. Strong bots win 31% of the tricks they play it in.
- 7 Treasure (+0.17 off-suit, +0.27 as trump). Whoever wins the trick scores a point for each 7 in it. Strong bots win 59% of the tricks they play it in.
- 11 Monarch (+0.08 off-suit, -0.09 as trump). When led, the opponent must play the 1 or their highest card of that suit. Strong bots win 65% of the tricks they play it in.
- 5 Woodcutter (-0.09 off-suit, -0.04 as trump). Draw a card, then put any card on the bottom of the deck. Strong bots win 38% of the tricks they play it in.
- 1 Swan (-0.22 off-suit, -0.16 as trump). Lose the trick with it and you lead the next one. Strong bots win 5% of the tricks they play it in.
- 9 Witch (-0.26 off-suit, -0.12 as trump). Counts as trump when it is the only 9 in the trick. Strong bots win 96% of the tricks they play it in.
At the table
- Keep the Fox until you know which trump suit you want, then use it to switch.
- Don't celebrate a hand full of 10s and 11s. Count how many tricks you can lose, too.
- Treat the plain mid cards (4, 6, 8) as liabilities, not as safe cards.
Which plan to aim for with these cards, humble or victorious, is a question of its own: we tested that too.
How we tested
Two searching bots played 20,000 rounds against each other. Before every card they imagine 24 ways the hidden cards could be dealt, play each option to the end of the round, and pick the best one on average. For every starting hand we recorded the cards in it and the final point difference, then fitted how much each card adds compared with an average card. A separate run of 4,200 rounds logged when each special card was played.
Card values depend on how well the bots use the cards. The bots decide Fox swaps and Woodcutter discards with simple rules rather than full search, so if anything the Fox and the Woodcutter are undersold here. We also split the self-play data in two halves and fitted each separately: both put the Fox first, the 2 and the 7 next, and the 6 last. More on how we test.
Image credits
- Photo: Renegade Game Studios (publisher image): The Fox in the Forest box with its cards fanned out around it
- Box art: Foxtrot Games / Renegade Game Studios (publisher image): The Fox in the Forest box, 3D view
