Spades Strategy: How to Bid and Play, Tested Over 10,000 Hands

The best Spades strategy is to count your tricks card by card and bid that count rounded down. Never waste a high card: when you can't win a trick, or your partner already has it, play your lowest card, and when you do win one, win it as cheaply as you can. Once your own bid is safe, keep taking tricks while the other team can still be set, and only bid Nil on a hand with no aces, no kings and no spade above the 10.

We tested each of these by having the computer players from Poozee's Spades play the same 10,000 hands over and over, changing one habit at a time. The winners are now how our computer plays: its old habits lost 15.24 points per hand to them. Here's how we tested, what won, and a few popular tips that barely mattered. New to the game? Start with how to play Spades for the rules and scoring.

How we tested

All four players used the rules of thumb our Medium computer plays by. Your team (you and your partner) then changed a single habit, while the other team kept the standard ones. We dealt the same 10,000 hands twice, once with your team on the standard habits and once with the change, so the luck of the deal cancels out.

We measured your team's margin: its points per hand minus the other team's. Bags count against you at 10 points each, because every 10 bags cost 100. With the standard habits, the team was set on 9.8% of its bids and took 1.22 bags per hand.

A computer playing a habit isn't a person, and your opponents won't play exactly like ours. But the results show which habits win when everything else is equal.

How to bid in Spades: count every card

Counting every ace and king as a trick is a common shortcut. It's roughly right for aces and badly wrong for a king on its own or in a long suit. We counted how often each kind of card took a trick, and rounded that to the nearest quarter of a trick. These are the values to add up.

Side suits (clubs, diamonds and hearts). "A suit of 3" means three cards of that suit in your hand, counting the card itself.

Spades, by how many spades you hold in all:

Two things stand out. A king is worth the most with one or two cards beside it, which let you wait until the ace has been played. A king alone usually falls to the ace; it's right on the line between 0 and 1/4, so we count it as nothing. And long suits lose value: the longer your side suit, the sooner someone else runs out of it and trumps your winners. Long spades work the other way, because once the other players run out of spades, your little ones win.

A worked example

Here's a hand from our test games:

• Clubs: ace, queen, jack, 9
• Diamonds: king, 8
• Spades: ace, 10, 5, 4
• Hearts: jack, 8, 2

The ace of spades counts 1, and with four spades, the 10, 5 and 4 count 1/2 each. The ace of clubs counts 1 and the queen of clubs, in a suit of 4, counts 1/4. The king of diamonds, in a suit of 2, counts 3/4. The rest count nothing. That's 1 + 1 1/2 + 1 + 1/4 + 3/4 = 4 1/2, so bid 4. In that deal, this hand bid 4 and took exactly 4, and the team bid 6 and took exactly 6. Rounded to the nearest trick, it would have bid 5: the team would have bid 7, taken 6 and been set.

How good is the count?

No count is exact, because the cards in the other hands matter too. Here's how many tricks players took on average, by their count rounded to the nearest trick:

Counts of 3 to 5 landed within a tenth of a trick on average. Lower counts took a little more than they counted, and counts of 6 or more a little less.

Pro tip: Short side suits make your low spades better, because you can trump with them. With three spades or fewer, here's how often a low spade took a trick, by the length of your shortest side suit: a void: 74%, one card: 58%, two: 35%, three or more: 17%. The count doesn't include this, so on average, players with a void took 0.70 tricks more than they counted, with a singleton 0.46 more, and the rest +0.07. So with a void, your hand is usually worth more than it counts.

Bid your count, rounded down

The count is right on average, so why round down? Because missing is much worse than going over. A set costs your team 10 points for every trick it bid, while an extra trick is a bag that costs 9. And good opponents keep taking tricks to set you (see habit 3 below), so a bid that's just barely makeable often isn't.

Your team counted with the table above. You changed how you turned your count into a bid, while your partner kept rounding down (a count that's already a whole number stays as it is). The other team kept the standard habits. "Set" is the share of your team's bids that fell short, "Bags" is bags per hand, and "Margin" is the change in your team's margin per hand.

Going a whole trick lower is the classic "play it safe" move, and it does cut your sets. But the extra bags cost more than the sets you avoid. Bidding high is far worse: being set more than half the time is a disaster.

Playing habits, ranked

Here's every playing habit we tested, with the same columns as above. "Margin" is the change in your team's margin per hand compared with the standard habits, so -1.00 means one point per hand worse. Each habit is explained below the table.

1. Never waste a high card

When you can't win a trick, or your partner already has it, play your lowest card. Our computer used to do the opposite: it threw away the highest card that would still lose, to get rid of it. That sounds tidy, but it means dropping your king under your partner's ace, when that king could have won a trick of its own later. Throwing your highest loser cost 15.94 points per hand and nearly tripled the team's sets, the biggest single mistake we found apart from not trumping at all.

The same goes for winning: when you're trying to win a trick and someone plays after you, win it with the lowest card that beats everything played so far, not your highest. Playing your highest winner cost 1.12 points per hand.

2. Cash your winners, and trump what you need

When you lead and your team still needs tricks, lead your sure winners first: the highest card left in a suit. Skipping that cost 10.91 points per hand: a winner you sit on can get trumped once someone runs out of its suit.

And when you're out of the suit led, your team still needs tricks and your partner isn't already winning, trump it. Holding back cost 22.54 points per hand, the biggest loss in the table.

Don't try to beat your partner, though. When your partner is winning the trick with a sure winner or a spade, or you're the last to play, overtaking your partner cost 2.78 points per hand.

3. Bags: what they really cost, and when to set instead

Each bag scores 1 point, but every tenth one costs 100, so a bag is really worth -9. With the standard habits, a team took 1.22 bags per hand, a 100-point penalty about every 8 hands. That's the usual reason to duck once your bid is made.

But it's better to keep taking tricks while the other team can still be set, meaning they still need tricks and there are enough left to take. Ducking as soon as your own bid was made cost 2.16 points per hand. Going further and taking every trick you could, even when the opponents were already safe or already set, cost 0.70, because those bags buy nothing.

(We priced every bag the same. With 8 or 9 bags already, or near the end of a game, a bag can cost more or less than that.)

4. Protect your partner's Nil, and play your own Nil right

With the standard habits, Nil bids succeeded 62% of the time. The partner of a Nil bidder plays to take tricks away from them: if the Nil bidder is winning the trick, beat their card, and if you play before them, play high. When partners stopped covering, the success rate fell to 41%, and the team lost 2.56 points per hand.

If you bid Nil yourself, get rid of your dangerous cards early: whenever you have cards that would lose the trick, play the highest of them. Playing your lowest instead saves the high cards for later, when they can win tricks you don't want: only 41% of Nil bids succeeded.

When to bid Nil

Nil is worth +100 if you take no tricks and -100 if you take even one. So it needs to succeed half the time just to break even, and more than that to beat the points a normal bid would have scored. We looked for hands that pass four checks:

1. No ace or king in any side suit.
2. No spade above the 10.
3. Three spades or fewer.
4. Every short side suit (one to three cards) has a card below the 9.

We went through 200,000 deals and played every hand that passed, and every hand that failed exactly one check, twice: once bidding Nil and once bidding normally. The partner bid normally both times and covered the Nil. "Nil gained" is how much bidding Nil changed the team's margin per hand, compared with bidding normally.

Hands that pass every check made Nil 64% of the time and gained the most points. One flaw is enough to turn Nil into a losing bid: every hand that failed a single check made it 46% of the time or less and did worse than bidding normally. The ace of spades can never lose a trick, so a Nil with it always fails.

The second row is part of the first: passing hands that still count to a trick or more. Those broke about even, so with one of them, Nil and a normal bid are a coin flip.

Tips that barely mattered

• The exact card values. Our computer used to bid with a more complicated formula. Rounded the way it used to be, that formula scored -0.20 points per hand against it, within the normal ups and downs of 10,000 deals (about 0.7 points either way). Rounding down mattered much more than the exact values.
• Leading your short suit. It's common advice, so you can trump that suit sooner. It helped a little (+1.36 points per hand), but that's close enough to the noise that we left our computer leading from its longest suit.
• Ducking high after your bid is made. Playing low until your bid is made and then throwing high cards to dodge bags cost 0.42 points per hand. Just duck low all hand.
• Bidding Nil at all. Bidding Nil on the right hands helped, but only by 1.48 points per hand, because just 2.6% of hands pass the checklist.

Practice against the computer

On Poozee's Spades, Medium plays by the standard habits above. Easy makes frequent random mistakes. Hard goes further: it imagines many possible deals of the cards it can't see, plays each option out with Medium's habits and picks the best. Hint suggests Medium's bid, or the card Hard would play.

If you also play Hearts, our Hearts strategy guide is tested the same way.

Keep playing

• Play Spades with a computer partner
• Hearts strategy, tested over 20,000 hands
• Play Hearts against three computer players
• FreeCell strategy
• Gin Rummy strategy: when to knock, tested the same way