You know the moment. Partner opens, you have a fit, a couple of aces and a singleton, and somewhere around the four level a little voice asks whether there are twelve tricks in this hand. Then a second voice answers: four spades is safe, the slam might go down, and nobody ever got glared at for stopping in game.

That second voice is wrong more often than you'd think. Not always. But deciding when to bid a slam is less about courage and more about arithmetic, and the arithmetic is kinder to slam bidders than most club players assume.

The bonus math at IMPs

Start with team scoring, where the numbers are cleanest. Say you have a spade fit and 6♠ makes if one finesse works. Non-vulnerable, bidding and making the slam scores 980. Stopping in 4♠ and taking twelve tricks scores 480. That's a 500-point difference, worth 11 IMPs.

Now flip it. The finesse loses, 6♠ goes one down for minus 50, and the pair who stopped in game collects 450. Another 500-point swing. Another 11 IMPs.

Same size win, same size loss. So the break-even point for a small slam at IMPs sits right around 50 percent. Vulnerable, the numbers get bigger (1430 against 680 when it makes, minus 100 against 650 when it doesn't) but the swings still mirror each other at 13 IMPs each way. If you've looked at how the IMP scale turns score differences into IMPs, this is exactly why team players call a small slam a coin flip.

A coin flip. Sit with that for a second. Any slam that hinges on a single finesse is, mathematically, a perfectly respectable contract. Most players treat it as reckless.

It's the same logic that pushes good teams into thin vulnerable games, which we covered in when to bid game and when to settle for a partscore. The bonus is big enough that you don't need the odds on your side, just not against you.

Grand slams are a different animal

The grand slam is where caution earns its keep. Bid 7♠ vulnerable and make it, and you score 2210 against 1460 for the pair in 6♠ with an overtrick. That's 13 IMPs. Go one down, though, and you're comparing minus 100 to plus 1460, a 17 IMP loss.

Risking 17 to win 13 pushes the break-even up to roughly 56 or 57 percent. And honestly, most strong players want a grand to be a lot better than that, because real hands are rarely as tidy as "it needs one finesse." Did you count the tricks right? Is there a ruff the defense can take at trick one? A grand that's "probably cold" has a funny way of meeting the one 5-0 split in the deck.

My rule: bid the grand when you can count thirteen tricks, not when you can imagine them.

Matchpoints change the question

At matchpoints the size of the bonus barely matters. What matters is what everyone else in the room does with the same cards, which is the whole point of how strategy shifts between IMPs and matchpoints.

Picture a 50 percent slam. If the entire field stops in game, bidding it gets you a clean top when it makes and a clean bottom when it fails. Over a long run that works out to, well, average. You took a lot of risk to stand still.

Now picture a 70 percent slam that half the room is bidding. Stopping in game means you lose to every one of those pairs whenever it makes, which is most of the time. Conservative bidding feels safe at matchpoints. It quietly bleeds you, one board at a time, and you never see the wound.

So the matchpoint version goes something like this: bid slams that are clearly better than even, and don't agonize over the marginal ones unless you suspect the field is already there.

Why so many players underbid slams

Mostly it's memory. A failed slam is vivid. You remember the bad trump split, partner's sigh, the silence while someone wrote down the score. A slam you didn't bid barely registers, because at a home game nobody tells you it was there. You wrote "4♠ +2," someone said "nice hand," and the deal went back in the shuffle.

That asymmetry is poison. One painful failure gets replayed for months while a dozen missed 60 percent slams vanish without a trace. No wonder people drift conservative.

A better habit, when you're sitting in game with extra values: ask how many tricks you can actually count, what single thing has to go right for the twelfth, and what the odds of that one thing are. A 3-2 trump break with five missing is about 68 percent. A finesse is 50. Two finesses where you only need one of them is about 75. Those three numbers do most of the work at the table.

Finding out whether you were right

The hardest part of improving your slam judgment is that you so rarely learn the answer. At a club, the traveller shows you that five other pairs bid six while you sat in four. At home there's no traveller, so the missed slam simply disappears. This is where playing the same pre-dealt boards as other tables earns its keep: with Bridge@Home, your foursome's result on each deal is scored against everyone else who held those exact cards, so a missed slam shows up as a real number instead of a vague "we probably should have bid on."

That number stings a bit. It should.

The takeaway

Small slams at IMPs need roughly even odds, no more. Grands need clearly better than that. At matchpoints, think about what the room is doing before you decide. And whatever the scoring, stop letting the one slam that went three down last spring make every decision for you. Count the tricks, then trust the count.