untpoker
Contents / Chapter 08
Chapter 08
POKER FROM FIRST PRINCIPLES

Preflop strategy from first principles

8.1 The state vector

Before the flop, record at least: your two cards, your position, all prior action, the positions of remaining players, effective stack depth, blind and ante structure, and any relevant opponent tendencies. In tournaments, payout context may also matter. A preflop chart is a compressed policy for a particular assumed state distribution; change the assumptions and the chart changes.

8.2 Effective stack

Only chips that can actually be won from an opponent matter for the maximum loss or gain in a heads-up pot. If you have 120 big blinds and your only opponent has 45, the effective stack is 45 big blinds. Deep stacks create more postflop optionality; short stacks make preflop all-ins comparatively more important.

Stack geometry

Effective stack = min(your stack, opponent stack) [heads-up pot]

8.3 Why position widens opening ranges

Opening from early position risks a re-raise or call from many players. Opening from the button leaves only two players behind and buys postflop position if called by a blind. Even with identical cards, the branch-weighted EV changes. That is why "Is A-9 suited playable?" is an incomplete question. "Playable from which seat, at which stack depth, against which response distribution?" is the right question.

8.4 Raise, call, or fold

Calling preserves more combinations in the pot at a lower immediate price but often sacrifices initiative and may invite additional players. Raising can win immediately, isolate opponents, build a pot with strong hands, or generate fold equity with hands that are not favorites when called. Folding has zero incremental EV and can be optimal even after you have posted a blind or already invested chips earlier.

8.5 A shove toy model

Suppose the pot before your action is P0. You shove an effective amount S. Villain calls with probability q. When called, your equity is e. Ignoring side pots and rake:

Heads-up shove model

EV(shove) = (1-q)*P0 + q*[ e*(P0 + S) - (1-e)*S ]

This model separates three ingredients: dead money P0, call frequency q, and rangeversus-range equity e. The MIT preflop lecture uses exactly this style of modeling: estimate how a pushing range performs against a calling range, then search for stable responses or approximate equilibrium behavior.

8.6 Nash thinking without worshipping a chart

In short-stack games, action can collapse toward shove-or-fold because a large raise commits most of the stack anyway. In such toy states it is often possible to search for a pair of push and call ranges where neither player benefits from unilateral deviation. That is a Nash equilibrium for the simplified game. It is not automatically the best exploitative strategy against a particular human, and it does not transfer unchanged to deeper stacks, different antes, different positions, or modern rake structures.

Historical-course caution

The MIT lecture contains specific 2015-era push/call approximations and bankroll heuristics. This book keeps the modeling method but does not present those numerical ranges as timeless current strategy.

8.7 Practical beginner policy

  • Start tighter in early position and wider in late position.
  • Prefer entering pots with a raise rather than passive limping when your hand is strong enough to play.
  • Respect re-raises more when stacks are deep and opponents are tight; respect stack commitment more when stacks are short.
  • Measure stack depth in big blinds and ask what future streets remain meaningful after a raise.
  • Treat every static chart as a hypothesis generated under assumptions, not as a substitute for EV reasoning.

UNTPOKER · AN INDEPENDENT EDUCATIONAL GUIDE