Introduction
Folding AK on a tournament bubble is one of the most stressful situations in poker. Most recreational players fold automatically. So do many regulars. But is it actually correct?
In this article we analyze a real spot with specific stacks, positions, and action. We apply GTO logic, calculate equity, factor in ICM, and solve the spot with a solver to give you a definitive answer you can apply in your next session. If you're coming from pot odds calculation, this is the next level: combining pure math with tournament pressure.
Auto-folding AK on the bubble isn't conservative. It's costly. Every time you fold AK in a +EV spot, you're giving away equity that other players capture.
Discussion
The spot: real tournament context
Let's analyze a typical mid-to-high stakes online tournament spot that plays out hundreds of times a day at tables all over the world. If you're still learning tournament dynamics, we recommend reviewing the Introduction to MTT course first before continuing.
— Tournament: Online MTT, 1,500 players, 46 remaining. — Bubble: 46/45 get paid. — Blinds: 1,000 / 2,000 ante 200. — Hero (MP2): 28bb effective, opens 2.5bb with AKo. — BTN: 18bb effective, shoves all-in. — Blinds: fold, fold. — Pot if Hero calls: 40bb total.
The central question: Hero opens from MP2 to 2.5bb with AKo. BTN shoves 18bb. The blinds fold. Hero must decide whether to call the all-in or fold.
Why this spot is critical
We're on the tournament bubble. 46 players remain, 45 get paid. One more elimination unlocks money for everyone. ICM pressure is at its peak. Every all-in or fold decision has direct consequences for prize equity, not just chip counts.
Hero is a mid-stack with 28bb. Not a desperate short stack, not a comfortable chip leader. He's in the zone where doubling up puts him in contention, and busting leaves him empty-handed.
The decision: chip EV vs. dollar EV (ICM)
To answer correctly we need to compare two perspectives: the purely mathematical chip-based one (chip EV) and the real tournament money perspective (dollar EV under ICM). You can practice these types of calculations with the integrated ICM calculator in the tournament hub.
Chip EV: the mathematical foundation
If we ignore the bubble and treat chips as linear money, the calculation is straightforward. Hero needs to call 18bb to win a pot of 2.5 + 1 + 0.5 + 18 = 22bb (his own open, the big blind, the small blind, and BTN's shove). After his call, the total pot would be 40bb.
Equity needed to break even: 18 / 40 = 45%. AKo against a typical BTN shove range (22+, AQs+, AKo, some bluffs like A5s–A2s) has approximately 48–52% equity depending on the exact range. Against a tighter range (TT+, AQs+, AKo) it drops to 43–45%.
Chip EV conclusion: against a wide shove range (very common from BTN vs. MP2 with 18bb), AKo has enough equity to call. Against a tight premium range, the call is marginal or slightly -EV.
Dollar EV: the ICM filter
Now we apply the Independent Chip Model. ICM converts chip stacks into real money equity based on the prize structure. Near the bubble, chips have a non-linear value: losing all of them is more costly than winning the pot is beneficial.
With a typical 15/12/10/8/7… online tournament structure, the ICM calculator shows that folding AKo in this spot has a slightly higher dollar expected value than calling against a premium range. However, against a standard BTN shove range (which includes medium pairs, AXs, and some suited connectors), the call remains +EV in dollars despite the bubble.
The key point: BTN's range matters more than the bubble in this specific spot. If BTN is shoving a wide range — which is GTO with 18bb from BTN against an MP2 open of 2.5bb — Hero cannot fold AKo.
Why it matters in the ecosystem
The common mistake: auto-folding out of bubble fear
The most frequent error in this spot isn't mathematical. It's psychological. Players see "bubble" and their brain translates it as "survive at all costs." That survival instinct ignores three critical factors:
— Hero is not a short stack. With 28bb, folding doesn't guarantee survival. The blinds keep going up and Hero will gradually bleed pot and fold equity. — BTN is shoving a wide range. With 18bb from BTN, a GTO shove includes many hands that AKo dominates (AJo, KQo, 66–88, Axs). Folding AKo against that range is leaving money on the table. — ICM doesn't always say fold. The ICM calculator confirms that against a standard BTN range, calling has higher prize EV than folding. Only against extremely tight ranges (JJ+, AKs) is folding correct under ICM.
The auto-fold fallacy: many players believe folding is the safe play on the bubble. But folding isn't free. It costs fold equity, it costs a double-up opportunity, and it costs the pressure you take on by being left with 10bb after paying blinds.
GTO solution: what the solver says, step by step
To solve the spot precisely, we set up the tree in a standard GTO solver with the tournament parameters. You can replicate this exact analysis in the GTO Lab or with the real engine in Solver Pro. Here are the steps and conclusions.
Step 1 — Set up the solver tree. Parameters: 8 players at the table, stacks matching the real spot, blinds 1,000/2,000 ante 200, tournament prize structure loaded. Hero in MP2 with AKo. BTN with an optimized shove range for 18bb effective vs. an MP2 open.
Step 2 — Analyze the equilibrium solution. The solver assigns a call frequency to each of Hero's hands based on its equity against BTN's shove range. AKo receives a 100% call frequency in the pure GTO solution (chip EV). When we apply the ICM factor with the prize structure, AKo still receives a call in 85–95% of combinations depending on the exact stack sizes of players at other tables.
Step 3 — Interpret the mixed frequencies. In GTO, mixed frequencies (call 90%, fold 10%) don't signal indecision. They mean the hand is at the indifference threshold: both lines have similar EV. For a human player, the optimal simplification is to call AKo 100% of the time. The small fold component in the mixed solution exists so Hero isn't predictable, but in practice the EV difference between calling 100% and calling 90% is negligible.
Step 4 — Adjust for reads on the opponent. If you have data showing that BTN is an extremely tight player who only shoves QQ+ and AKs on the bubble, then folding AKo is correct. This is an exploitative deviation, not GTO. To dive deeper into these kinds of adjustments, check out the Exploitative Play and Reads course. GTO tells you the strategy that can't be exploited; exploitative play tells you how to take advantage of your opponent's mistakes.
Additional context
Comparison table: decision by BTN range
— 3% range (QQ+, AKs) → AKo equity ~38% → fold (nit range, premium only). — 8% range (TT+, AQs+, AKo) → AKo equity ~43% → marginal fold (tight but callable without ICM). — 15% standard GTO range → AKo equity ~48% → call (optimal 18bb BTN range). — 25% wide range → AKo equity ~52% → clear call (includes KQo, AJo, 66+). — 40% extreme range → AKo equity ~55% → instant call (maniac, lots of bluffs).
The table makes it clear that only against ultra-premium ranges (3% or less) is folding the correct line. Against any wide or standard shove range from BTN, AKo has enough equity to call even under ICM pressure.
5 quick decisions by context
— Call if BTN is shoving a standard GTO range for 18bb from BTN (22+, A2s+, AJo+, KQs, KQo, QJs). AKo has 48–52% equity and the ICM pressure is not severe enough to turn a +EV call into a fold. — Fold if you have a solid read that BTN only shoves JJ+ or QQ+ and AKs. In that case your equity drops below the ICM threshold and folding is correct. — Fold if you are a critical short stack with fewer than 10bb and another player at the table is about to bust. In that context, temporary survival has extreme value. — Call if your stack is medium-to-large (20bb+) and you need a double-up to be competitive at the final table. Folding leaves you in a slow bleed situation from the blinds. — When in doubt, measure. Loading the exact spot into the GTO Lab takes 30 seconds and eliminates the guesswork.
Conclusion
Conclusion: when to fold and when to call AK on the bubble
The definitive answer is: it depends on your opponent's shove range, not simply on the fact that you're on the bubble. Don't let the word "bubble" dictate your decisions without math to back them up. Use the ICM calculator, know your opponent's range, and make decisions based on real equity, not fear.
From theory to practice
Train this spot with professional tools. Load the pre-configured spot in our GTO Lab with the exact stacks, blinds, and positions, run the solver, and compare your decision to the GTO solution. Repeat the spot with stack and range variations: the system gives you instant feedback and tracks your improvement.
When you're ready to take your tournament game to the next level, the course catalog covers everything from MTT fundamentals to advanced strategy under ICM pressure. For full access to the solver, lab, and guided learning paths, check out the Pro and Elite plans or start with the recommended learning path.
Calculate your own ICM bubble spot
If you're interested in solving these types of spots with your own real numbers, we've published a free ICM calculator that applies the same math (Malmuth-Harville + Nash) used in this analysis. It's free, no sign-up required, and returns $EV(push) vs. $EV(fold) with ICM pressure already factored in. Useful both for reviewing hands from your history and for training decisions ahead of your next tournament.