Bags in Spades: What They Really Cost, Tested Over 10,000 Games

Bags in Spades are the extra tricks your team takes beyond its bid. Each bag scores 1 point, but every time your team's count reaches 10 bags, you lose 100 points and the count drops by 10. Over a game that averages out to roughly 10 points per bag, but the real cost of one more bag swings from nothing at all to the full 100, depending on how many bags you have and how close the game is to its end.

We measured that with the Medium computer players from Poozee's Spades, over whole games to 500. The rules themselves are in how to play Spades, and bidding and play hand by hand are in our Spades strategy guide. This guide is about bags across a whole game: when they hurt, when they don't, and what to do about them.

How bags work

This is exactly how Poozee scores them:

• Your bid and your partner's are added into one team bid. Every trick your team takes beyond it is a bag, and still scores 1 point.
• A trick taken by a player who bid Nil is always a bag, even though it doesn't count toward their partner's bid.
• A team that falls short of its bid (is set) takes no bags for that bid.
• Each team's bag count carries over from hand to hand, next to its score. When a hand brings the count to 10 or more, the team loses 100 points and the count drops by 10, keeping whatever is left over.
• A hand that brings the count to 20 costs you twice. A team on 9 bags that bids 2 and takes all 13 tricks adds 11 bags, making 20, so it scores 20 + 11 - 200 = -169. That's rare, but possible.

Some tables play bags differently, with a different limit or penalty, or no bag penalty at all, so agree on it before the first deal. Everything below uses Poozee's rules.

Worked example: a bag count over five hands

Here's one team's bag count, starting from zero. "Bid" is your bid + North's, "Took" is the tricks you and North took, "Bags" is the bags taken that hand, and "Total" is the bag count after the hand. West and East bid 3 each and take the remaining tricks.

1. Hand 1: the team bid 5 and took 7, so it scores 50 + 2 and takes 2 bags.
2. Hand 2: it took 5 tricks for a bid of 6, so it's set: -60 and no bags.
3. Hand 3: your Nil fails (-100) and North makes 4 with one over (+41). Your trick didn't help North, but it's still a bag, so the team takes 2.
4. Hand 4: bid 4, took 8: +44 and 4 bags, for a count of 8.
5. Hand 5: bid 3, took 6. That's 3 more bags, and 8 + 3 = 11 crosses 10, so 33 for the hand minus 100 is -67. The team carries 1 bag into the next hand.

How much bags matter over a whole game

We played 10,000 games between two teams of Medium computer players. A game lasted 11.8 hands on average, and each team took 14.2 bags per game, about 1.2 per hand. A team paid at least one 100-point penalty in 78% of games and two or more in 14%.

Bags also decide games. Our Medium computer players never look at the score or the bag count, so every hand of a game plays out the same whatever happened before it. That let us replay each game with the same hands and different scoring. With no bag penalties for either team, the other team would have won 15.4% of games. The losing team paid at least one bag penalty in 81% of games, and if it had been let off its own penalties, the losing team would have won 34% of them.

What one more bag really costs

The strategy guide prices a bag at 10 points, a tenth of a penalty. That's right on average, but in penalties no single bag costs 10 points: it costs the full 100 or nothing, and which one depends on the rest of the game.

Here's why. If your team will take more bags later anyway, one extra bag now only makes a penalty you were going to pay arrive a hand or two sooner. It costs you 100 points only when, without it, you'd have finished the game one bag short of your next penalty.

So we went back through the 10,000 games, and at every hand where a team made its bid, we gave it one more overtrick and replayed the rest of the game: 212,587 times in all. The table shows how often that one bag led to one more 100-point penalty by the end of the game. "Bags" is the team's count after the hand, not counting the extra one. "Early" is a hand that began with both teams under 400 points, "Late" one that began with either team at 400 or more, and "Last hand" the hand that ended the game (Early and Late leave it out).

Averaged over all of them, one bag led to one more penalty 11.7% of the time, close to the usual one in ten. But look at how it changes:

• Early in a game, your count barely matters. One more bag led to a penalty between 9% and 14% of the time, whatever the count. Even the 10th bag isn't a disaster early on: at 9 bags, one more led to an extra penalty only 11% of the time, because you would usually have taken that 10th bag a hand or two later.
• Late in a game, high counts get expensive. At 7, 8 or 9 bags, one more bag cost a penalty 22% to 28% of the time, against 11% to 14% early on.
• Late in a game, low counts are nearly free. With 0 to 4 bags, there's rarely time left to reach 10: 4% or less.
• In the hand that ends the game, only the 10th bag matters, and it always costs the full 100. At 9 bags, that one bag turned a win into a loss 16% of the time.

Pro tip: Late in a game, glance at both bag counts before an overtrick. With 9 bags in what's likely the last hand, one more trick costs your team 100 points, so it's the one time a bag really is a disaster.

Should you stop setting the other team when your bags are high?

The strategy guide found that once your bid is made, it pays to keep taking tricks while the other team can still be set. Since bags late in a game cost more, we tested ducking instead, late in a game (either team at 400 or more), once the team's bags reached a given count. That changes fewer games, so we tested it over 40,000 games instead of 10,000.

It didn't help. Ducking late at any count won 0.8 percentage points fewer games. Ducking only from 7 bags scored -0.04 percentage points, within the noise, and at 9 bags it cost 0.2 percentage points. So keep playing to set the opponents, even late in a game.

How to avoid bags, and what it costs

Every way of taking fewer bags gives something up. Here's what a few habits did to your team (you and North both changed habits, and the other team kept the standard ones), per game. "Bags" is bags per game, "Penalty" is 100-point bag penalties per game, "Sets" is the hands per game in which your team fell short of its bid, and "Won" is the share of games your team won.

The standard habits are the ones from the strategy guide: duck with your lowest card, and once your bid is made, keep taking tricks only while the other team can still be set.

Ducking as soon as your bid is made cut your bag penalties from 0.93 to 0.65 per game, but your team won 4.1 percentage points fewer games. The cost came from somewhere else: your team set the other team 1.15 times per game, and ducking dropped that to 0.67. Throwing your highest losing card dodges even more bags, but it gives away the high cards you need to make your bid: sets more than tripled.

Ducking low until your bid is made and then dumping high cards made no real difference (about 0.5 percentage points either way). So the best way to avoid bags is to bid accurately in the first place, not to throw tricks away.

Pro tip: If you only remember one rule, make it this one: duck with your lowest card. Your low cards can't win tricks later, and your high cards can.

Bid one trick lower or one trick higher?

When your hand is a trick either way, which mistake is cheaper? Here only you bid differently, and your partner and the other team keep counting their tricks and rounding down. "One trick lower" and "higher" are one trick off that standard bid.

Bidding a trick low is the cheaper mistake, but both are expensive: one trick lower more than doubled your bag penalties, and one trick higher more than tripled your sets. Bidding your count rounded down beat both.

Should you bid higher when you're close to 10 bags?

A common idea is to bid up when your team is near a penalty, so you won't take more bags. We tested it: whenever your team already had a given number of bags at the start of a hand, both of you rounded your count to the nearest trick instead of down.

Even at 9 bags, bidding up lost 2.3 percentage points, and the earlier a team started, the worse it did. Bidding a full trick higher at 8 or more bags won only 28% of games. A set costs your whole bid, and that's worse than a bag even at 9.

Sandbagging: does it pay?

"Sandbagging" means two things at Spades tables. The first is bidding low on purpose and collecting overtricks. That's the "One trick lower" row above: it won 31.1% of games instead of 49.9%.

The second is feeding the other team bags. Once both teams have made their bids, every trick left is a bag for whoever takes it, so you try to lose them all: when you can't win a trick anyway, throw your highest card instead of your lowest, so you have fewer winners left for later. We tested that too, with you and North sandbagging whenever both bids were made and no Nil was still alive.

It worked, but only barely: +0.04 points per hand over 10,000 hands (bags priced at 10 points each), and 0.11 percentage points more games won. It changed the winner of only 17 of the 10,000 games. Most likely that's because by the time both bids are safe, few tricks are left and you rarely have a choice of cards that matters.

Keep playing

• Play Spades and watch the bag count next to each team's score
• Spades strategy: how to count your tricks, bid, and when Nil pays
• How to play Spades, with every scoring rule and worked examples
• Euchre strategy, another partnership game tested the same way