Heads-up push/fold charts
The equilibrium for the shove-or-fold endgame, from 2 to 25 big blinds. Our engine solved it, the trainer grades you with it, and every number on these pages comes out of the same file, which you can download and check.
One chart for every depth
Each cell is the deepest effective stack, in big blinds, from which that hand still shoves (left) or calls a shove (right), checking every depth from 2 bb up. Your stack is at or below the number: play it. Above: fold. 25+ means the hand plays at every depth we solved; a dash means it folds even at 2 bb. Hover or tap a cell for the detail. An asterisk marks a hand whose range is not one clean band (see below).
Small blind: shove up to
deepest stack, in bb, the hand shoves from
Big blind: call up to
deepest stack, in bb, the hand calls a shove from
Range size by depth
Shove and call percentages are weighted by real combos (a pair is 6 of the 1,326 starting hands, a suited hand 4, an offsuit hand 12) and by the equilibrium frequency of mixed hands. Open a depth for both full charts, the copyable range text and the EV of all 169 hands.
| Effective stack | SB shoves | BB calls | Chart |
|---|---|---|---|
| 2 bb | 89.8% | 100.0% | 2 bb chart → |
| 3 bb | 77.9% | 92.3% | 3 bb chart → |
| 4 bb | 73.8% | 73.7% | 4 bb chart → |
| 5 bb | 71.6% | 61.8% | 5 bb chart → |
| 6 bb | 68.7% | 54.4% | 6 bb chart → |
| 7 bb | 66.5% | 48.7% | 7 bb chart → |
| 8 bb | 61.8% | 45.1% | 8 bb chart → |
| 9 bb | 60.6% | 40.7% | 9 bb chart → |
| 10 bb | 58.3% | 37.3% | 10 bb chart → |
| 11 bb | 54.5% | 35.1% | 11 bb chart → |
| 12 bb | 53.5% | 33.0% | 12 bb chart → |
| 13 bb | 51.5% | 31.0% | 13 bb chart → |
| 14 bb | 48.3% | 29.3% | 14 bb chart → |
| 15 bb | 45.8% | 28.6% | 15 bb chart → |
| 16 bb | 44.8% | 26.4% | 16 bb chart → |
| 17 bb | 42.9% | 25.0% | 17 bb chart → |
| 18 bb | 42.2% | 23.7% | 18 bb chart → |
| 19 bb | 41.0% | 22.6% | 19 bb chart → |
| 20 bb | 40.2% | 21.7% | 20 bb chart → |
| 21 bb | 38.1% | 20.8% | 21 bb chart → |
| 22 bb | 38.2% | 20.0% | 22 bb chart → |
| 23 bb | 38.0% | 19.0% | 23 bb chart → |
| 24 bb | 38.0% | 17.9% | 24 bb chart → |
| 25 bb | 36.3% | 17.4% | 25 bb chart → |
Where the chart is not one clean band
8 small-blind hands shove at some depth, stop, and shove again deeper. We print them rather than smooth them out. The last column prices the gap: across it, shoving and folding are never more than that many big blinds apart. The largest is 0.104 bb, small next to the hands that are clearly in or out, but not nothing. Fictitious play converges to an equilibrium, not to a tidy picture of one, and at near-indifferent hands the two are not the same thing.
| Hand | Every depth up to | Plays again at | Max |EV| in gap (bb) |
|---|---|---|---|
| Q5s* | 19 bb | 22–23 bb | 0.005 |
| J5s* | 14 bb | 16 bb | 0.004 |
| T6s* | 20 bb | 22–24 bb | 0.013 |
| 84s* | 7 bb | 10 bb | 0.008 |
| 63s* | 2 bb | 6–7 bb | 0.064 |
| 53s* | 2 bb | 5–12 bb | 0.070 |
| 43s* | 2 bb | 5–7, 10 bb | 0.104 |
| Q9o* | 20 bb | 24 bb | 0.019 |
Every hand in the calling chart is monotone: once it stops, it stays stopped.
When push/fold is the right frame
Shove-or-fold is a model of the game, not the game. With very short stacks it is close to exact, because a min-raise commits so much of the stack that folding to a shove is rarely an option anyway. As stacks get deeper, a real small blind also limps and min-raises, and those options are worth something. This table cannot tell you how much, because it was not allowed to use them: deeper than about 25 bb we stop rather than pretend.
What it does tell you exactly is the price of a shove at a given depth when the other player defends correctly, and the price of calling one. That is the arithmetic every short-stacked decision in a sit-and-go or tournament endgame starts from.
Knowing the chart and playing it under a clock are different skills. The trainer deals real hands, at the frequency they really arrive, and prices each wrong answer in big blinds.
Drill push/fold →- Heads-up, blinds 0.5 / 1, no ante. Stacks are the effective stack in big blinds.
- The small blind may only shove or fold; the big blind may only call or fold. No limps, no min-raises.
- Solved by fictitious play over a 169×169 all-in equity matrix, 60,000 sampled runouts per matchup, card removal corrected. Every line ends all-in or folded, so there is no postflop model and no realisation fudge.
- EV is the aggressive action minus folding, in big blinds, against the opponent's equilibrium range. Frequencies between 20% and 80% count as mixed, the same cut-offs the trainer grades with.
- Raw data: pushfold.json (the exact file the trainer and these pages read).