A short stack is a situation, not a chart.

After-play study, not an at-table aid. Fictional examples explain how to define and compare a spot. No push/fold ranges are supplied or claimed to be solved.

“I had eight blinds” is the beginning of a question.

Two players can hold the same cards with the same nominal stack and face different decisions. One is first into an unopened pot on the button. The other is in the big blind after a raise and a call. The number of players, posted money, available actions, other stacks and tournament prizes can all differ. A chart cell is useful only after its situation matches yours.

This is a study method for the small-stack decisions that are easy to compress into slogans. It also gives a reviewer a way to explain why two charts disagree without immediately declaring one wrong. Begin with position and stack terminology if effective stacks are unfamiliar.

Record a state, not a stack label.

Write the blind level, ante type, seats dealt in, button location, action so far and stack convention. “8 BB at the start of the hand before posting” is different from “8 BB behind after posting.” Use the software or chart's own convention when comparing results, and retain your raw chip counts so that a conversion can be checked.

Effective stack is opponent-specific. A button with 12 BB facing blinds with 7 BB and 30 BB has a different amount at risk against each. Reducing this to “7 BB effective” hides the deeper opponent. In a multiway all-in, different contribution levels can also create side pots.

A big-blind ante is posted once per hand.

At a six-handed table with a 0.5 BB small blind, a 1 BB big blind and a 1 BB big-blind ante, the initial pot is 2.5 BB. Do not multiply that big-blind ante by six. With a 0.1 BB ante from each of six players instead, the initial pot is 2.1 BB.

At unchanged blinds, a player who completes an ordinary full six-hand orbit pays a total of 2.5 BB in the first structure or 2.1 BB in the second, assuming every obligation is posted in full. That is bookkeeping, not a survival clock: blinds can rise, the table can change and the player may win or lose chips before completing an orbit.

Keep actual forced contributions in the pot ledger. A blind is a live contribution and an ante is not a bet toward the amount to call. This distinction matters most when very little is left behind.

Do not mix these four families.

Choose the decision family before consulting a reference

Family	What must match
First-in decision	All earlier players folded. Count everyone still to act and preserve their stacks.
Facing an open	The opener's size and assumed range, plus players behind, are part of the state.
Calling an all-in	The actual amount owed and the shover's range matter. You cannot make the existing shover fold by calling.
Limped or multiway pot	Prior contributions and additional opponents create a different game from an unopened heads-up comparison.

“Push/fold” is an action abstraction: a model removes alternatives such as limping or a smaller raise from a particular decision. It does not prove those alternatives are inferior in the unrestricted game. Even a very short stack can have legal non-all-in options in some situations. State what the model allowed instead of turning a simplifying assumption into a universal rule.

Give every chart a small compatibility label.

Before comparing a hand recommendation, record the table size and position convention; all relevant stacks; blind and ante structure; action history; chip-value or prize-value objective; permitted actions; and the source or version. For a prize-sensitive comparison, keep the payouts and remaining-field information too.

Then label the match as exact for the recorded inputs, approximate with named differences, or not comparable. “Exact” here describes matching inputs, not a claim that a program has solved the game perfectly. A screenshot with missing settings is not an exact reference.

For example, a six-max, no-ante, first-in 5 BB chip-value chart is not an exact answer for a five-handed 4.5 BB spot with a big-blind ante near a payout jump. It may be educational, but the differences are the point of the study rather than inconvenient footnotes.

A nearby depth is not a license to interpolate.

A binary shove/fold label at 4 BB and another at 5 BB do not by themselves specify the answer at 4.5 BB. A reported mixed frequency is also not a confidence score. Prefer an exact matching calculation when available; otherwise retain the mismatch and describe the result as an approximation. Do not silently round a difficult spot into certainty.

A small comparison beats a wall of charts.

Choose one family, such as a button first-in decision. Save one complete input record and predict which factors should matter before looking at a result. Compare one change at a time: ante structure, an opponent's stack, or the payout objective. Record what changed and what remained fixed.

Finish with a statement you could explain without the chart: “This comparison changes the posted pot but keeps the action tree and stacks fixed,” or “This difference might be due to the payout model, so the chip-value reference is not enough.” A study session that identifies a mismatch has produced something useful even before it produces an action recommendation.

For a limited-energy session, reconstruct one hand and stop before querying a tool. For a longer session, add a matching reference and one sensitivity check. There is no need to memorize dozens of isolated cells before understanding the question they answer.

Download the short-stack input record