From Hold'em to Omaha

The ceiling fell out

Here is the whole mental reset in four numbers. Every one of them was computed by this app's own engine — none of them is quoted from a book.

• Hold'em, A♠A♥ all in against one random hand: 85.2%.
• Omaha, A♥A♦K♥K♦ — about as good as a starting hand gets — against one random four-card hand: 70.7%.
• Omaha, ordinary aces A♥A♦8♣3♠ against a random hand: 61.6%.
• Omaha, that same premium A♥A♦K♥K♦ against a hand that will actually call you — a connected J♣T♣9♠8♠: 61.5%.

Read the last two lines together. Your best hand against a real opponent is worth about what your worst aces are worth against nobody in particular. That is the game.

Method: the first number is our own five-card evaluator run over every seven-choose-five combination, 1000000 simulated deals, standard error 0.04 points. The Omaha percentages against a random hand are 1000000 simulations each, standard error 0.05 points. The last one is not a simulation at all: it is an exact enumeration of all 1086008 possible boards (全枚举真值，并与 phevaluator 对拍过).

One pair is no longer a hand

Deal two random starting hands, run the board out, and ask what the winning hand was. 1000000 times, both games:

• Hold'em: 43.1% of those showdowns are won by one pair or worse. A straight or better wins only 18.3% of them.
• Omaha: one pair or worse wins 10.7%. A straight or better wins 42.9%.

The two games are almost mirror images. Top pair with a good kicker, the hand you have been building pots with for years, is now roughly what a high card used to be.

Do not build a pot with one pair.

Multiway pots, medium-to-deep stacks, and any board that is not bone dry. The more players and the more money behind, the harder this holds.

Aces are still the best preflop hand in the game — A♥A♦K♥K♦ is 70.7% against a random hand. That is a pair, and it is not the exception that proves the rule; it shows what the rule is really about. Aces do not win by making one pair at showdown, they win by being ahead before the flop and by letting you shrink the stack-to-pot ratio while you are still ahead. Heads-up, on a dry board, at a low SPR, one pair of aces is frequently the whole hand. What loses money is paying off three streets with it in a four-way pot.

Method note, stated so you read the number correctly: this is two random hands taken to the river, not a distribution of real showdowns (in real hands people fold). Both games measured the same way, so the comparison holds even though the absolute level does not describe your table.

Everybody's hand is nearly the same size

We sampled 2000 random Omaha starting hands and measured each one against a random opponent (20000 simulations per hand). The spread:

• The 5th percentile hand is 40.6%, the median is 49.9%, the 95th percentile is 62.0%. That is a band 21.4 points wide.
• Same sampling on hold'em hands (10000 simulations each): 35.6% / 49.7% / 65.1%, a band 29.5 points wide.

Nine out of ten Omaha hands live inside a twenty-point window. That single fact drives almost everything else in this course: preflop edges are small, most pots are decided after the flop, and postflop is where the money is.

Do not expect to win preflop. Take small edges, keep position, and play the flop.

Around 100 big blinds, in normal single-raised or three-bet pots where there is still money behind on the flop.

Twice this is false. First: when the pot is already built. After the four-bet in C2 you have 125 chips behind against a pot of similar size — there is barely a postflop left, so the preflop equity is the hand. Second: "small differences" is not "no differences". 2♣2♦3♥3♠ is 38.6% against a random hand, well below that 5th percentile, and our reference ranges fold it from every seat. Close together does not mean interchangeable.

What to take away

Three habits to unlearn, each with the number that kills it:

• "My aces are a monster." 85.2% became 70.7% at best, 61.6% for the ordinary version, 61.5% against a hand that actually calls.
• "Top pair is a hand." 43.1% of showdowns used to go to one pair or worse; here it is 10.7%.
• "I can pick a spot and run people over preflop." Nine hands in ten sit inside a 21.4-point band. There is nobody to run over.

What replaces them: your draws get bigger than anything you have seen (a four-card hand can flop a wrap, a straight draw with far more outs than an open-ender — that is C14 to C17), the nut hand matters more than the strong hand, and position gets more valuable, not less. That is the rest of this camp.