The most common tournament mistakes, and what the solver does instead

Three years ago we published a three-part series about preflop mistakes in tournaments, built from a training session Vlada Stojanovic gave for Poker Academy. The ideas were good; the write-up was not. Stack depths went missing, the format was never stated, and a few sentences do not survive a look at the solver. So I have put the whole series back together in one piece, with every spot re-opened in the app.

Everything below is chipEV, not ICM: standard tournament simulations, 9-handed with 1bb of total ante in the middle (the 80bb spot is 8-handed). Percentages are shares of all 1,326 starting combinations, the way the app counts them. Each section names its node, so you can open the same chart yourself, and the last one shows what ICM does to the numbers.

1. Jamming the big blind too often against a small blind open

Blind versus blind, 20bb effective. The small blind raises to 3bb. You are in the big blind with 1bb already posted and 1bb of ante beside it, so there is 5bb in the middle and it costs you 2bb to call.

The solver folds 32.0% of hands here, calls 54.5% and moves all in with 13.3%. The jam is real, and it is the small part of the strategy. Most players use it as their main one.

The old article explained it like this: the small blind’s range is “highly polarized, indicating a strong inclination towards calling”. That chain does not connect. The question is what the small blind continues with when you jam, and the next node answers it. Facing the all-in it folds most of what it opened and calls with only 37% of its opening range: AA down to 99, AK to A9 suited, A7s, A6s, KQs, KJs, KTs, JTs, J9s, T9s and KJo. A third of those combos hold an ace.

Now look at your own hand. KJo, QTs and T9s are beaten by most of that list, so putting 20bb in against it is the worst use of them. In a blind battle you are also the one in position after the flop. Offsuit aces are the opposite: a third of that calling range holds an ace, so one in your hand removes a tenth of it, and the hand plays badly after the flop anyway. It would rather have the 5bb now.

• Jam: 22–66, most offsuit aces (AKo to A9o, A4o, A3o, A2o; A8o and A6o part of the time), AKs, AQs, AJs, A2s.
• Call: everything with playability, so 77 and up, A5s to A3s, suited broadways, suited connectors, offsuit broadways. A7o and A5o call as well.
• Fold: the bottom 32%, almost all of it offsuit.

Short stacks get their own article: short stack strategy from 10 to 20bb.

2. Treating a limped blind battle like a normal pot

Same blind battle, 17.5bb, but now the small blind limps. The pot is 3bb (1bb limp, your 1bb, 1bb ante) and it costs you nothing to see a flop. The app does not even draw a fold button at this node.

The original article said the big blind can “check and fully realize their equity before the flop”. Equity gets realized after the flop, not before it, and how much you keep depends on position, stack depth and the board. The simpler version: checking is free, so weak hands keep their equity instead of throwing it away.

The next claim in the old article was that the small blind limps about 54% of its range, and that “a significant portion of their range will fold to a raise”. It does not. Against the 2.5bb raise, the limping range continues 81% of the time: it calls with 37.2% of all hands and jams another 6.6%, out of the 53.9% it limped. You are not raising these people off their hands. You raise for value, for the dead money, and to charge the hands that flop badly.

The rest of Vlada’s advice holds up: build the raising range so you are not ashamed of it when the limp-jam comes back. The solver’s raise is mostly offsuit (K8o, Q8o, J7o, T7o, 97o, 95o, 74o), sitting next to the strong part: AA through 99 and the suited broadways. Suited hands with playability, like J7s, 54s and 92s, just check.

And the “diagonal rule” from the old post turns out to be measurable. Split the grid along the pairs: suited hands sit above it, offsuit hands below. Of the suited combos the big blind raises, 82% call when the small blind jams over the top. Of the offsuit combos it raises, only 17% do.

More on the blinds: big blind basics and blind-versus-blind play.

3. Opening the wrong hands from the small blind

Still 17.5bb, everyone folds to you in the small blind. The solver plays 84.3% of hands: limp 53.9%, raise to 3bb 15.4%, all in 15.0%, fold 15.7%. The mistake sits in which hands you raise.

The raising range is not the top of the range. AA, AKs, AQs, KK, QQ, JJ and TT raise, and so do A9s to A6s, J9s, T9s, T8s, K8o, Q8o and Q7o. Meanwhile AJs, ATs, A5s to A2s, K9s, Q9s, AKo and AQo all limp, and 22–55, the offsuit aces from AJo down, 98s, 87s and 76s go all in.

Vlada’s rule for finding them was to ask which hands make a bad limp-call. A8s limped is a hand that flops top pair with a weak kicker and very little else; raised, it wins the pot outright often enough to be worth the same. Two caveats. The solver is nearly indifferent for most of these hands: A8s is worth 1.62bb raising and 1.62bb limping. And it is the shape of the range that does the work here, not the individual hand. A raise made only of premiums is easy to play against; one with no premiums in it is worse.

4. Defending too wide four-way, and never jamming

35bb effective, 9-handed. The hijack opens to 2.3bb, the button calls, the small blind calls, and you are in the big blind. There is 8.9bb in the middle (2.3 × 3, your 1bb, 1bb ante) and it costs 1.3bb to call.

Wonderful odds, and still not enough for a lot of hands: the solver folds 47.0% here. Take the old article’s example, J8o. Calling is worth −1.13bb and folding −1.00bb, so folding wins by 0.13bb even at 12.7%. Three players are already in, all of them hold hands that dominate J8o, and you play the rest out of position.

The play almost nobody makes is the other one: the solver moves all in for 35bb with 10.6% of hands. The lecture called them “J9s, QTs, all suited broadways”. J9s is a suited connector, not a broadway hand, but the range is right: suited broadways and suited connectors down to T9s, the suited aces, every pair from 77 to KK, and AKo, AQo, AJo, ATo. Aces always raise to 12.8bb instead, and QJs and AJs mostly do, which is the tell. The jam here is the second tier of the range, not the top of it.

It works because the hijack opened 29.1% of hands and can call the jam with under 30% of that: AA down to 77, AKs to ATs, AKo, AQo, AJo, KQs, KJs. Nine big blinds of dead money, folded to you seven times out of ten, beats playing a three-way pot out of position.

5. Missing the squeeze

80bb effective, 8-handed. The cutoff opens to 2.5bb with 35.5% of hands, the button calls, the small blind folds, and you look down at A♠︎T♠︎ in the big blind. There is 7.5bb in the middle and 1.5bb to call.

The solver squeezes to 12.3bb with 10.1% of hands, and ATs is one of them, 100% of the time. The lecture said ATs “is ahead of the 35% CO opening range by a lot”, which is worth putting a number on. Against exactly the range the solver opens from the cutoff in this simulation, one suit combination of ATs has 58.6% equity (Monte Carlo, 200,000 hands, so read that as 58.6% ± 0.1). So ATs is ahead of that range, and I would not call it by a lot.

The edge is small. Raising ATs is worth 1.10bb, flatting 1.07bb: three hundredths of a big blind. One hand like that is no disaster, and it is worth fixing because the spot comes up fifty times a session. AJs is clearer: 1.77bb raising against 1.68bb calling.

The other half of the reason is what the cutoff has to do next. Facing the squeeze it continues with 30% of its opening range and no more: 14% calls, 12% moves all in, 4% 4-bets. Seven times out of ten the player who opened is done with the hand, and the button still has to survive the squeeze behind him.

6. Not 3-betting and not jamming over wide late-position opens

35bb, the button opens to 2.3bb with 48.2% of hands, call it half, and you are in the big blind for 1.3bb more into a 4.8bb pot, a price of 21.3%. The solver folds 21.8%, calls 59.4%, 3-bets to 8.6bb with 10.9% and moves all in for 35bb with a further 7.9%. Part three of the old series quoted “13.7% raises” from a chart nobody could see; here the aggressive part of the range is 18.8%, split between two sizes.

The two aggressive ranges do different jobs, and not the ones people expect. The 3-bet is the top: AA, KK, QQ, JJ, TT, AKs–ATs, AKo, KQs, JTs. The jam is the awkward middle: 22 through 99, A5o, AJo, ATo, AQo, KQo, K7s. Small pairs are the clearest case: 22 is worth +0.15bb jamming, −0.01bb calling and −0.15bb 3-betting.

Why 22 or K7s can shove 35bb into a 4.8bb pot: the button folds 74.9% of its opening range to it. That is the 73% from the lecture, and it holds up. Half the combos it does call with hold an ace, so one in your hand deletes an eighth of them. That is why A5o (76% of the time) and K7s (100%) jam, while J8s, T8s and 98s flat, in this simulation and against this size.

One wording to correct: A5o and A2o at 35bb are not an “all-in open with BTN”, as the old part three put it. You are not opening; you are 3-bet jamming over the button. A5o does it 76% of the time, and A2o is genuinely indifferent: jam, 3-bet and call are all worth −0.26bb.

The deeper version of this spot has its own article: is 40bb too deep to 3-bet all in?

All of this is chipEV

Vlada put his own caveat on all of it: “this is Chip EV and as soon as we introduce some ICM, there will be less jamming.” He is right. Chip EV assumes a chip won is worth what a chip lost costs. In a tournament it is not, and the app’s ICM simulations carry the gap as a bubble factor. At 35bb with 20 players left and everyone already in the money it is 1.358; with 50 left and 30 paid it is 1.387. A chip you lose costs you about 1.36 to 1.39 times what a chip you win gains.

You can see it in the ranges. The same big-blind spot against a button open at 35bb jams 5.3% and 5.0% in those ICM sims, where the chipEV chart jams 7.9%. They are 8-handed with a 2bb open against 9-handed and 2.3bb, so read it as a direction, not an exact difference: under ICM the calls stay and the all-ins shrink. Bounties move the maths in the opposite direction, which is PKO, and it has its own rules.

What to take from this

• Blind versus blind at 20bb: the jam is 13% of your range, not your default.
• Limped pots: checking is free, so the big blind never folds. The raise is for value rather than folds, because the limper continues 81% of the time.
• Small blind opens: the raising range is not the top of your range. It is the hands that make bad limp-calls, mixed with enough strong ones.
• Four-way pots: 12.7% pot odds still folds 47% of hands, and the play you are missing is the all-in.
• Squeezes: small edges, taken often. The player who opened folds 70% of the time.
• Against a 50% button open: 3-bet your top, jam the awkward middle, stop calling with small pairs.

None of these is a chart to memorize. Each one is the same question about the other player: what does he continue with, and how does my hand do against that rather than against the range he opened? Open the node, look at what he has to do next, and the right action usually names itself.

Cheers! 🙂