40bb 3-bet all-in from the small blind: chipEV vs ICM ranges

40 big blinds feels like a lot of room. You can open, 3-bet small, flat and still play poker after the flop. So moving all-in with 40bb over a single open sounds reckless. The solver disagrees, a little. It does jam 40bb from the small blind, with a thin slice of hands, and which slice depends on who opened and on how the tournament pays.

I pulled the spots straight from the Poker Academy preflop solutions. ChipEV: CE-Symmetric, 9-handed MTT, everyone 40bb, 1bb big-blind ante, 2.3bb opens. ICM: three 9-handed 40bb sims with a classic ante (0.12bb per player, 1.08bb in total) and 2.25bb opens, at 100, 50 and 20 players left. Percentages are shares of all 1,326 starting hands unless I say otherwise.

The maths of a 40bb jam

The CO opens to 2.3bb and the button folds. The pot is 2.3 + 0.5 + 1 + 1 = 4.8bb. If you jam from the small blind, you put in 39.5bb more to win those 4.8bb when everyone folds.

In the chipEV sim the big blind folds 93.4% of hands to that jam and the CO folds 78.3% of its opening range. Together that’s 0.934 × 0.783 = 73.1% (ignoring card removal), well short of 89.2%. So a 40bb jam is never a pure bluff. It needs hands that do fine when they get called, and it has to take equity away from hands that would otherwise see a flop.

SB vs CO: 1.8% jams

Against a CO open the small blind folds 72.6%, calls 14.4%, 3-bets to 8bb with 11.3% and jams 1.8% of hands. The jam is not the top of the range: AA, KK and QQ 3-bet small.

The hands that jam, with their all-in frequency:

• Always: JTs (100%).
• Most of the time: AQo (71%).
• Part of the time: 88 (44%), KTs (42%), KJs (32%).
• A little: QJs 14%, AJo 13%, KQo 12%, QTs 11%, AKo 8%, 99 6%.

My reading, as a heuristic: these hands have decent equity against a calling range but are awkward after a small 3-bet, facing a 4-bet shove or a big pot out of position. JTs is the only pure jam here.

How the CO answers the jam

The CO is getting a price: 37.7bb to call into a 44.3bb pot.

It still folds most of its opening range. The CO opens 36.8% of hands at 40bb and calls the jam with 21.7% of that range: AA–77 (66 half the time), AKs–ATs, AKo, AQo, KQs, KJs, AJo about half the time and KTs a little. Everything else folds. The app shows it on the node itself, including the pot odds.

SB vs BTN: 4.1% jams

Against a button open the small blind jams more than twice as often: 4.1% of hands, with 13.0% 3-bets to 8bb, 11.6% calls and 71.4% folds.

• Always: AQo, QTs, JTs (100%), 77 (96%).
• Most of the time: KTs, AJo (74%), KQo (55%), 88 (52%).
• Part of the time: 44 (42%), A5s (35%), 55 and QJs (24%).

The opener is what widens it, not the stack: that is 40bb in both spots. The button opens 50.6% of hands and the CO 36.8%, so the button’s range is wider and weaker. It folds a similar share of it to the jam (79.0% against 78.3%) and calls with 21.0% of it. My reading: the button’s calls come out of a weaker opening range, so more medium pairs and broadways have enough to jam.

What ICM does to the same spot

ChipEV counts chips. ICM counts prize money, and in a tournament the chips you lose are worth more than the chips you win. The ICM sims here use a 30-place payout: 1st gets 19.2% of the prize pool, 2nd 13.8%, 3rd 10.38%, down to 1.85% for 10th to 13th and 1.12% for 24th to 30th. Each has the whole table at 40bb; the rest of the field has other stacks, and the average across all players left is 35bb. The app labels them by field size left: 50% (100 players, 70 from the money), 25% (50 players, 20 from the money) and 10% (20 players, already in the money).

40bb SB, all-in / 3-bet small / call / fold	vs CO open	vs BTN open
ChipEV	1.8 / 11.3 / 14.4 / 72.6	4.1 / 13.0 / 11.6 / 71.4
ICM, 100 left	0.2 / 11.0 / 13.6 / 75.3	1.2 / 12.9 / 15.2 / 70.7
ICM, 50 left	1.3 / 10.0 / 9.1 / 79.5	3.0 / 12.2 / 9.6 / 75.2
ICM, 20 left	– / 9.5 / 14.0 / 76.5	– / 11.6 / 17.2 / 71.1

Percent of all hands. A dash means the tree has no all-in option in that sim.

In both ICM sims that offer the jam it comes out smaller than in chipEV, and it doesn’t shrink in a straight line: with 50 left it’s bigger than with 100 left.

20 left: the chipEV jams become calls

In the money, with 20 players left, the solution has no small blind all-in at all. One honest note: this sim’s tree only gives the small blind fold, call and a 3-bet to 7.6bb, so the solver never tested the jam and turned it down. What you can read from it is what happens to the hands that jammed in chipEV:

• JTs, KJs and KTs call 100%.
• 88 calls 93% and 3-bets 7%.
• AQo calls 75% and 3-bets 25%.

Against the button it’s the same picture: JTs, QTs, KTs, 88, 77 and 44 call every time, while AQo, AJo and KQo split between a call and a 3-bet to 7.6bb. An earlier version of this post said the all-ins “switch to almost pure calls”. For the CO spot that holds, in the same sim. It just wasn’t the solver turning the jam down.

50 left: the callers tighten more than the jammer

With 50 players left, 20 from the money, the jam is back: 1.3% against the CO and 3.0% against the button. That’s more than at 100 left. The reason sits on the other side of the table.

Near the bubble the CO calls the jam with only 13.0% of its opening range, against 21.7% in chipEV: AA–TT, AKs, AQs, AKo and AQo most of the time. The big blind folds 96.2%. With both players folding, the jam wins the pot right away 83.7% of the time, against 73.1% in chipEV.

In ICM the small blind loses more prize equity when it gets called, but it gets called much less. Against the button that trade is good enough for a clear jam range:

• Always: A4s, A2s, KJs, KTs, AQo (100%), A3s (95%), QJs (92%), A5s (86%).
• Mostly gone: the pairs (44 jams 14%). 77 and 88 now call, and JTs and QTs call too.

The jam range moves from pairs and connected broadways towards suited aces. My reading, as a heuristic: an ace blocks the calling range (AA, AK, AQ), and a hand like A4s still has some equity when it does get called. A small pair would rather see a cheap flop than put its tournament life on a flip.

With 100 left the jam is small: 0.2% against the CO (KJs 33%, AQo 10%) and 1.2% against the button (KJs 78%, KTs 67%, AQo 50%, QTs 47%). The CO calls 17.3% of its opening range there, so the fold equity sits between chipEV and the bubble.

So, is 40bb too big to 3-bet all-in?

• No, but it’s a small play. 1.8% of hands against a CO open and 4.1% against a button open in chipEV.
• It’s not your best hands. AA, KK and QQ 3-bet small. Jams are hands like JTs, AQo, KTs and medium pairs.
• The opener matters, the stack is the same. The button opens 50.6% and gets jammed more.
• ICM shrinks the jam at 100 and 50 left, and the 20-left sim doesn’t offer it at all.
• Near the bubble, watch the caller. When openers call a jam much tighter, the jam comes back with suited aces.

None of this is a rule to memorise hand by hand. The mix changes with the ante, the open size and the payouts, so look up the spot that matches your tournament.

Cheers! 🙂