Position and Opening Ranges

Why position is worth more here

This page is an argument, not a measurement, and it is labelled that way on purpose. What we measured is in C3; what we conclude from it is below, and you are entitled to weigh it yourself.

The measured inputs: nine hands in ten sit inside a 21.4-point equity band, and one pair or worse wins only 10.7% of showdowns. Follow the chain:

• Small preflop edges mean more hands are correct to play, so fewer pots end before the flop.
• More hands playing means more pots are multiway.
• Hands that need three cards to come in also need more streets, so more money goes in on later streets.
• Every one of those later decisions is made better by acting last.

If you come from hold'em you already believe position is valuable. The claim of this course is that it is worth more here, not less — and the reason is that the pot is decided later, more often, with more people in it.

Open wider the later you act.

Six-handed cash, around 100 big blinds, in an unopened pot — the situation our reference ranges cover, and nothing else.

The small blind breaks it. Our reference ranges open the small blind wider than the cutoff, because only one player is left to get through. But the small blind acts first on every postflop street, which is the worst seat at the table once the flop comes. Reading "wide range" as "good position" is how players end up playing the small blind like a button. A wide opening range there is a statement about how many players are left, not about how good your seat is.

Same hand, different seat

Four hands, walked across the table. These are the actions our reference ranges give — qualitative actions only, which is all a curated table is entitled to publish.

• J♠T♥9♠2♥ — double-suited, but with a 2♥ hanging off the bottom. Folded under the gun, in middle position and in the cutoff. Opened on the button and in the small blind.
• J♠T♠9♥8♦ — a clean single-suited rundown. Folded under the gun, mixed in middle position, opened from the cutoff and the button. This is what a boundary hand looks like: the same hand crosses the threshold one seat at a time.
• 2♣2♦3♥3♠ — folded everywhere, including the button and the small blind. Position does not rescue everything.
• A♥A♦8♣3♠ — opened from every seat, under the gun included, despite being the bad aces from C5.

Two lessons sit in that list. The first is that position is a dial, not a switch: hands do not go from unplayable to premium, they cross a line somewhere in the middle of the table. The second is that the dial has stops at both ends — the strongest hands open everywhere and the weakest fold everywhere, and no amount of position changes either.

A mixed entry means both actions are defensible at that seat. It is not "pick randomly" and it is not "we could not decide" — it is a hand sitting on the boundary, where the difference between opening and folding is small enough that other things (the table, the players behind you, how you play postflop) decide it.

What a reference range is, and what it is not

First, why there is no grid. Four cards from 52 is 270725 combinations, which our classifier folds into 16432 structurally distinct hands. Two cards from 52 is 1326 combinations and 169 distinct hands, which is exactly why hold'em can draw a 13×13 grid and Omaha cannot. Our range table is a category list instead: every one of the 16432 classes carries an action, and you can look up any four cards you like.

What it is. A reference range built in-house: six-handed cash, 100 big blinds, 5 opening seats plus big-blind defence against a button open. Calibrated against public total-entry frequencies, then adjusted by structure using our own equity engine. Every step is written down in docs/M2-SOURCES.md, and you can regenerate the whole table from source.

How wide it opens. This one we can put numbers on, because it is not an estimate — it is a property of the table you can count. Weight all 270725 hands by how many combinations each is, count a mixed entry as 0.5 of a hand, and you get this reference table's opening frequency: under the gun 17.6%, middle position 22.3%, cutoff 30.5%, button 48.6%, small blind 39.5%. The button opens nearly three times as wide as under the gun. That is what "open wider the later you act" looks like with a ruler against it.

What it is not. It is not a solver solution, and those percentages are not "correct frequencies" — they are the opening frequencies of this table. Its confidence tier is curated, which in this app means one specific thing: we do not publish per-hand mixed frequencies. "This hand raises such-and-such a share of the time" is a number we do not have and will not invent. The opening frequency of a whole table is a countable fact; the raising frequency of one hand is not. Do not read the first as if it were the second.

Use the table as your default. If it says fold, fold.

In the exact situation it covers, against opponents who are playing a normal game, while you are still learning the shapes.

It is a calibrated table, not a verdict. Three specific places it stops short: the mixed entries are genuinely two-way and either action is fine; it says nothing about three-bet pots, squeezes, limped pots, multiway pots or any stack depth other than 100 big blinds; and against opponents who are far from normal, the right answer moves and the table does not. Treat it as a starting point you can recompute, never as an authority you cannot argue with.

What to take away

• Position is a dial. Most hands are not "playable" or "unplayable"; they cross a line somewhere between under the gun and the button.
• A wide range is not a good seat. The small blind opens wide because few players are left, and then acts first for the rest of the hand.
• The reason position is worth more here is the 21.4-point equity band: pots go further, with more players, and acting last decides more of them.
• The table is a reference range, not a solution. All 16432 hand classes carry a qualitative action, no frequencies, and a written recipe you can rerun.

Look your own hands up — including the ones you folded last session, and the mixed ones you have been guessing at.