Hold em Math
Hold em Math
Hold em Math
Introduction
If you’re amoung the many people who lack a math gene and quake at the thought of
having to do math to learn poker, fear not; you don’t need to be a math whiz to understand the
game. All you need is some common sense, a basic understanding of the concepts underlying
poker math and the ability to apply those concepts at the tables.
These notes are basically conversations I’ve had with myself in an attempt to explain the basic
math concepts behind Texas Hold’em, why the concepts are important to playing winning poker
and when they need to be applied. I’m not a math whiz so don’t expect any long, involved
explanations of why the math works; if you’re looking for that try Brian Alpach’s pages, Wikipedia:
Poker Probability or MathWorld.
The following assumes you are familiar with Texas Hold’em and know the structure of the game.
If you’re totally new to the game and have no idea what Limit, Blinds, Flop or Position refer to you
1
may want to read some introductory material before proceeding.
1
Basic introductions to the game can be found at http://www.gamblingpress.com/casino-games/poker/texas-
holdem-rules.htm or http://www.ultimatebet.com/rules-strategy/texas-holdem.html
Let’s say you’re playing a standard Limit $1/$2 game; there are 10 players at the table. What is
2
your fair share of winning hands?
If the game is played with all cards face up and the winner is determined solely by who has the
best hand once all the cards are dealt, over the long run, you can expect to win 10% of the hands
dealt or one of every 10 hands dealt.
Of course, your wins are not likely to be evenly spread. You may lose 30, 100, 1000 hands in a
row or you may win 30, 100 or 1000 hands in a row. However, at the end of the day, after playing
thousands of hands, the number you win will be close to, if not exactly, 10% of all the hands
you’re dealt. And the same will hold true for every other player at the table
What happens if you and all the other players must make a bet on every hand? If you play 10,000
hands and make a $1 bet on each hand, the cost of playing is $10,000. Trouble is, you will only
win 10% of those hands. Is there a profit in this?
Each time you win you receive $10: $1 from each of the other 9 players plus your $1 original bet.
That works out to 1,000 hands (10,000 x 10%) where you win $10. 1,000 x $10 equals, you
guessed it, $10,000. Exactly the amount you spent to play the game. You break even! That’s no
fun. Who wants to spend all that time playing 10,000 hands only to break even?
What if you are allowed to decide when you will bet but all the other players must continue to bet
$1 on every hand? Let’s say you decide to make a bet only when you are dealt pocket rockets:
AA. Will you make a profit then?
First, you need to know how many times you can expect to be dealt AA. Luckily, this has already
been worked out for us. Mike Caro has posted the information in his The Probability of Being
Dealt Specific Hold’em Hands Before the Flop table. Turns out your chances of being dealt a pair
of Aces (or any pair, for that matter) are 0.45% or 220:1. Which means that for every 221 hands
dealt to you, one of those hands will be a pair of Aces.
So, if you play 10,000 games you can expect to get pocket Aces 45 times (10,000 / 221 or 10,000
x 0.45%). Assuming you bet $1 every time you are dealt pocket Aces and assuming you win with
those Aces every time, the game would now only cost you $45.00. How much would you win?
Well, if all the other players were still betting $1 every hand then you would expect to win 10 bets
on every hand you bet. That’s $1 from each of the other 9 players in the game plus your original
$1 bet, which you would get back. So, you’d win $450 ($10 x 45) for a net profit of $405 ($450 -
$45). Now, instead of breaking even after 10,000 hands you’d actually be up $360. Nice!
Notice that we are winners in this scenario not because we are great poker players but because
our opponents were forced to bet every hand!
Let’s modify our example slightly. Suppose we continue to bet $1 only when we are dealt Aces
and our opponents are forced to bet only half their hands instead of every hand. How does that
change our winnings? Ten thousand hands will still cost us $45. And, assuming our Aces always
2
For more on the concept of fair share hands see More Hold’Em Excellence: A Winner for Life by Lou
Krieger, ConjelCo, PA
hold up, we will now win $5 x 45 or $225. We win less but we still win considerably more than we
bet.
Of course, that’s not what happens in the real world, no one forces our opponents to bet every
hand. Nor is there only one betting round, there are four. But what also happens in the real world
is that our opponents not only (a) play too many hands, they also, (b) stay with them when there
is little chance their hand will win by the river. The net result is they cannot even hope to break
even, instead they simply lose money. We can win if
1. We choose only to play hands that have a positive expectation (good chance of winning),
and,
2. We play against people who play too many hands and who continue to play when their
hands have a negative expectation (when they can’t possibly win)
To choose winning hands we need to understand what a winning five card hand is. In other
words, we need to understand the poker Hand Rankings.
Hand Rankings
Most of us learned the hand rankings as children:
Royal Flush
Straight Flush
4 of a Kind
Full House
Flush
Straight
3 of a Kind
Two Pair
One Pair
High Card
What we may not know is just why the hands are ranked in the above order. Why is a Royal
3
Flush the best hand in poker? Why does a straight beat three of a kind?
4
The following table shows the number of ways each five card hand can be made when dealt
from a 52 card deck.
For the most part, the hands are ranked strictly according to how often they might be dealt; the
rarer the hand, the higher the ranking.
Note that there is a slight anomaly when choosing the best five cards from seven cards; you are
more likely to get one pair (43.8%) or two pair (23.5%) than you are to get a high card hand
(17.4%); yet a pair or two pair still beats a high card! Brian Alspach explains why the pair hands
are ranked higher:
“…you are basing the ranking on 5 cards so that if you were to rank a high card
hand higher than a hand with a single pair, people would choose to ignore the
pair in a 7-card hand with a single pair and call it a high card hand. This would
have the effect of creating the following distortion. There are 81,922,260 7-card
hands in the last two categories containing 5 cards which are high card hands. Of
these 81,922,260 hands, 58,627,800 also contain 5-card hands which have a
pair. Thus, the latter hands are more special and should be ranked higher (as
they indeed are)”
3
You may notice the table does not include the probability of a Royal Flush. That’s because it is essentially
a straight flush. In fact, there are 4,324 ways to make a Royal Flush or a 0.00323% chance you will be dealt
one when 7 cards are dealt.
4
The table is taken from http://www.rgpfaq.com/ranking-why-general.html For an explanation of the actual
math involved to create the table see http://www.math.sfu.ca/~alspach/comp20/
By which I think he means that all the one pair hands also include the high card hands.
Essentially there are 81,922,260 ways you can be dealt a high card hand (58,627,800 +
23,294,460). As there are 81.9k ways to make a high card and only 58.6k ways to make one pair,
the one pair hand is rarer and thus deserves to be ranked higher (this also explains why I’ve been
losing with my high card hands…there’s a combined 67.3% chance someone will have one or two
pair by the river and only a 17.4% chance my high card will be good. Had I only known this
earlier!).
Besides the mathematical probability involved in ranking hands there is also an arbitrary element.
At some point in the past people agreed that an Ace beats a King, a King beats a Queen, a
Queen beats a Jack, and so on. How they arrived at this high card scheme is not known. It’s not
based on any mathematical premise as there are the same number of Aces as there are Kings,
Queens, Sevens, Two’s or any other card in the deck. But it turns out that this arbitrary ranking
scheme is crucial in deciding which hand actually wins.
Regardless of the type of hand dealt: flush, straight, 3 of a kind, pair it is the hand with the highest
ranking card that wins! Three King’s beats three 7’s; a pair of 9’s beats a pair of 8’s; a straight to
the Jack beats a straight to the 6, etc.
There are two other factors to be considered in determining a hands rank: suitedness and
connectedness. These are based on mathematical theory. You are less likely to be dealt five
cards all in the same suit (3.02%) than you are to be dealt five connected (4.6%) cards.
5
These three hand attributes :
must all be taken into account when deciding which hands we will play. And, as Gary Carson
points out on p.145 of The Complete Book of Hold’Em Poker
Before we can jump into weighing starting hands we need to understand the math behind the
various starting hand guidelines.
5
Ed Miller discusses the value of these attributes on p.57 of Small Stakes Hold’em
Probability is the chance of a favourable outcome. Odds are probabilities re-stated as a ratio of
unfavourable to favourable outcomes.
For example, a coin flip can have one of two outcomes; the flip can result in the coin landing
heads-up or tails-up. If heads-up is considered to be the favourable outcome, we would say there
is a 50% chance (probability) that a coin will land heads-up when it’s flipped or we could say the
odds against a coin landing heads-up are 1:1. The odds are a ratio of one unfavourable outcome
(tails) to one favourable outcome (heads) given a universe of only two (1 + 1) possible
outcomes.
Since probability and odds are simply two different ways of expressing the same concept; you
can convert probability to odds and vice versa. There are two formulas that can be used to
convert probability to odds:
The first formula treats the probability as a percentage; the second formula treats the probability
as a fraction. Both formulas produce the same result – just use whichever is easier for you.
2:1 odds means we can expect 2 unfavourable outcomes to every favourable outcome for every
three (2+1) trials (attempts).
Note that the number of trials are implicit in the odds statements; if we add the two odd numbers
together we get the number of trials.
When using probability, all possible trials are represented by 100% (if we are using percentages)
or 1 (if we are using fractions). In other words, with odds, we are told how many trials are needed:
1:1, 2 trials; 2:1, 3 trials.
With probability we aren’t given the exact number of trials. Instead, probability represents the
portion of favourable outcomes with respect to all trials. Such that a 50% probability means ½ of
all the trials will be favourable, 33% probability means 1/3 of all the trials will be favourable, 25%
probability means ¼ of all the trials will be favourable and so on. Of course, the number of trials
is actually implicit since 50% = ½ implying 2 trials, 33% = 1/3, implying 3 trials, etc. It’s just not as
obvious at first glance.
To convert odds to probability rewrite the odds as a fraction and divide. For example,
5:2 = 2 / (5 + 2) = 2 / 7 = 28.57%
4:3 = 3 / (4 + 3) = 3 / 7 = 42.85%
To check whether you’ve done it correctly, convert your result back to odds.
Whenever we want to know what the odds of being dealt a particular starting hand, or flopping a
particular hand, or making a hand on the turn or river are we need to determine how the cards in
a standard deck can be combined or chosen from the available cards.
A standard deck of cards represents a limited, structured universe of objects. There are:
To find the number of unique two card combinations in a standard deck, consider:
13 x 13 = 169.
Therefore, there are a total of 169 unique 2 card combinations in a deck of 52 cards.
Which is why, if you have PokerTracker, you may have noticed, on the General Tab beside
Known Starting Hands, xx,xxx Hands – 169 Unique Hands. If you list out all the card
combinations, without regard to suit values, you’ll get 169 combinations.
A K Q J T 9 8 7 6 5 4 3 2
AA AK AQ AJ AT A9 A8 A7 A6 A5 A4 A3 A2
AK KK KQ KJ KT K9 K8 K7 K6 K5 K4 K3 K2
AQ KQ QQ QJ QT Q9 Q8 Q7 Q6 Q5 Q4 Q3 Q2
AJ KJ QJ JJ JT J9 J8 J7 J6 J5 J4 J3 J2
AT KT QT JT TT T9 T8 T7 T6 T5 T4 T3 T2
A9 K9 Q9 J9 T9 99 98 97 96 95 94 93 92
A8 K8 Q8 J8 T8 98 88 87 86 85 84 83 82
A7 K7 Q7 J7 T7 97 87 77 76 75 74 73 72
A6 K6 Q6 J6 T6 96 86 76 66 65 64 63 62
A5 K5 Q5 J5 T5 95 85 75 65 55 54 53 52
A4 K4 Q4 J4 T4 94 84 74 64 54 44 43 42
A3 K3 Q3 J3 T3 93 83 73 63 53 43 33 32
A2 K2 Q2 J2 T0 92 82 72 62 52 42 32 22
When determining how many unique 2 card combinations there were, we did not consider the suit
value of the cards. We assumed that any combination of AK was equal to any other combination
of AK and any combination of 5,3 was equal to any other combination of 5,3. In reality, this is not
true. The suit value of the cards does make a difference. We are not choosing 2 cards from 13
cards. We are choosing 2 cards from 52 cards (13 ranks x 4 suits).
C(52,2), where C = choose, 52 = all the available cards and 2 = the number of cards to
be chosen
A factorial is the product of all the positive integers from 1 to a given number. So 2! = 1 x
2. 52! = 1 x 2 x 3 x 4 … x 50 x 51 x 52.
Don’t worry, it looks more complicated than it really is. You don’t really have to multiply all 52
6
numbers! Since there’s only two cards being chosen, C(52,2) can be reduced to:
Therefore, when we consider the suit value of the cards, there are 1,326 possible 2 card starting
hands.
Multiplying 52 x 52 gives the number of ways you can combine the cards assuming you are
always free to pick and chose the cards you want to combine. But that’s not true with starting
hands, 1 card is removed, it has become distinct from the rest of the cards. It can only be
combined with a card pulled from the remaining deck. In effect you start with 52 cards, you pull
one, leaving 51 cards. You now pull 1 from 51. If you want another distinct two cards you pull 1
from 50, then 1 from 49, and so on. As you continue to pull two card hands the deck slowly
decreases until there are only 2 cards left. The Choose formula summarizes this behaviour into a
nice, neat formula.
Let’s say we want to know how many ways there are to make any pair; AA for example.
We know there are four Aces in the deck: A A A A . How many ways can you chose 2 of
these?
C(4,2) = (4 x 3) / (2 x 1) = 12 / 2 = 6.
There are 6 ways we can make a pair. This is fairly simple to prove. Here are all the possible
ways four Aces can be combined:
6
See http://mathforum.org/library/drmath/view/56106.html for an explanation of why this is so.
A A A A A A
A A A A
A A
Note that any pair can be made 6 ways; not just Aces.
How many ways can you choose any two specific cards?
Let’s say you want to know how many ways you can choose Ace, King.
There are 4 Aces and 4 Kings which can be combined with each other any way, suited or
unsuited. Answer: C(4,1) x C(4,1) = 4 x 4 = 16.
As with unique hands, we don’t care about the suit values of the cards. All we care about is the
number of Ace, King hands that can be created. And it’s easy to show that 16 is the correct
answer.
A K A K A K A K
A K A K A K A K
A K A K A K A K
A K A K A K A K
Count them up and there are 16 possible combinations. And once again, this is true for any
known two card combination, not just Ace, King.
How many ways can you choose two specific suited cards?
Let’s say you want to know how many ways you can choose a suited Ace, King.
There are 4 suits and only one Ace and one King in each suit, so what we are really doing is
choosing 1 suit out of four.
C(4,1) = (4 x 1) / (1 x 1) = 4 / 1 = 4
And once again, it’s easy to prove. These are the only available suited Ace, King combinations.
A K A K A K A K
How many ways can you choose two specific unsuited cards?
Let’s say you want to know how many ways you can choose an unsuited Ace, King.
We are choosing one Ace from 4 and one King from 3, since we don’t want to choose a King with
the same suit as the Ace.
Which makes sense since we said there were 16 Ace,King combinations, 4 of which are suited.
16 – 4 leaves 12 unsuited combinations.
What does all this have to do with knowing the odds you’ll be dealt a particular 2 card starting
hand?
Well, you now know how many possible starting hands there are ( 1,326 ) and how many ways
you can make a pair, two distinct cards or two distinct suited cards. Given that knowledge you
can work out the probability. Just divide the ways you can make the hand by the number of
possible starting hands.
For example, if you want to know the chances of being dealt a pair of Aces:
Or to get the odds, divide the possible starting hands by the ways you can make the hand.
Or any pair?
Or a suited Ace,King?
What about any two suited cards, if I don’t care what they are?
There are 4 suits, giving 4 possible suited hands for each 2 card choices.
4 x 78 = 312
1,326 / 312 = 4.25 (3.25:1) or 1 / 4.25 = 23.52% chance you’ll get two suited cards.
That’s fine if I’m the only person being dealt to but what about the 6, 8 or 9 other players at the
table? Won’t my odds of getting a specific hand change?
Nope. It’s true that we aren’t only dealing out 2 cards however you need to remember that odds
and probabilites are only best guesses. We have absolutely no idea what cards the other players
are receiving. Sure, we could guess that at least one Ace or two Kings have been dealt but we
don’t know that’s true. If we make those assumptions we will be piling up uncertainties, biasing
our guess either for or against specific hands. Why add bias? The minute we start assuming what
cards are in or out of the deck we favour certain combinations over others. If we only begin with
what we know at least all the uncertainties are equally balanced; we are not favouring one hand
combination over another.
It won’t, unless you’ve done some ground work, away from the table. Well, that’s not completely
true. If you’re heads up with KK you’ll know the odds are slim someone else will have KK or AK.
Why? Because you now know only 2 Kings were left in the the deck. There is only 1 way
someone else could make KK and only 8 ways (4 Aces x 2 Kings) someone else could make AK.
7
That’s 1,314 / 9 (8+1) = 146 (145:1) or 1/146 = 0.68% chance someone else has one of those
hands. Add 0.68% to the 0.45% someone has AA and there’s only a 1.13% chance our opponent
has a better hand.
But most of us couldn’t work that out in the 30 seconds we’ll have at the table. Which is why
knowing this stuff is really only useful if you work a lot of it out away from the table and file it away
in your memory for future reference. What it does really help with is figuring out what all those
VP$IP and PFR% numbers represent.
7
1,326 starting hands minus 6 ways for the KK we already made and 8 (16-8) ways to make AK which are
no longer possible.
If you’ve done some homework you’ll know he’s most likely raising a big pair or a big card
combination. Why?
What if their PFR% is 3%? Then they are probably only raising AA, KK, QQ and suited big cards.
Of course, you won’t know that for sure but if you haven’t seen any of their raising hands, it’s a
safe assumption. After awhile you may discover that the opponent with the 2.9% PFR is randomly
raising junk and not good hands at all. Or that the player with the 8% PFR also raises any suited
Ace. But if you don’t know that ahead of time using what you do know about the most likely
raising hands for someone with either PFR%’s can help you decide whether or not play you’re A,J
off-suit or pair of 77’s.
Another useful exercise is to work out what the various VP$IP’s represent. For example, the
common consensus is that a good player, in a full-ring game, will have a VP$IP of approximately
17%. Where does that number come from? Those playing big and medim pairs, big cards and
suited connectors.
Big Pairs >= 77: AA, KK, QQ, JJ, TT, 99, 88, 77 = 8 x 6 = 48 Hands
2 Cards > T: AK AQ AJ AT KQ KJ KT QJ QT JT = 10 x 16 = 160 Hands
Suited Connectors: T9 98 87 76 65 = 5 x 4 = 20 Hands
(48 + 160 + 20) / 1,326 = 228 / 1,326 = 17.2% VP$IP
Connectors: 54 43 32 = 3 x 16 = 48 Hands
Unsuited Connectors: T9 98 87 76 65 = 5 x 12 = 60 Hands
Low Pairs: 66 55 44 33 22 = 5 x 6 = 30 Hands
(228 + 48 + 60 + 30) / 1,326 = 366 / 1,326 = 27.6% VP$IP
What if they pay attention to position, adding hands as their position improves?
Let’s work out the number of hands they might play from MP and LP and add them to the 228
hands we worked out for the 17.2% VP$IP.
From MP:
Pairs: 66 55 = 2 x 6 ways = 12 Hands
Suited Aces: A9 A8 A7 A6 = 4 x 4 ways = 16 Hands
Suited Other: K9 Q9 Q8 J8 T8 98 = 6 x 4 ways = 20 Hands
Unsuited: KT QJ QT JT = 4 x 12 = 48 Hands
Total: 96 Hands played 2/3’s of the time = 64 Hands
From LP:
Pairs: 44 33 22 = 3 x 6 ways = 18 Hands
Suited: A5 A4 A3 A2 K8 K7 K6 K5 K4 K3 K2
J7 T7 97 96 87 76 76 75 65 54 = 21 x 4 ways = 84 Hands
Unsuited: A9 A8 A7 K9 Q9 J9 J8 T9 T8 98 97 87 = 12 x 12 ways = 144 Hands
Total: 246 Hands played 1/3 of the time = 48 Hands
Altogether:
(228 + 64 + 48) / 1,326 = 340 / 1,326 = 25.64% VP$IP
Which means anyone with a VP$IP greater than 26% is either playing all the usual hands from
any position or playing a lot of other hands.
That’s a total of 244 Hands / 1,326 possible starting hands = 18.4% VP$IP.
You can also see the stats by position for full-ring or 6 handed games.
For a 6 handed game, the following hands are +EV in any position or all but 1 position:
Contrary to common belief, loosening up in 6-max games may not be the wisest thing to do.
There is a noticeable drop-off in suited Ax hands (presumably because the pots are not big
enough to justify drawing to a flush). Pairs smaller than seven don’t fare well either. In fact, you’d
do better to add big unsuited cards rather than small pairs, suited Aces or suited Kings.
Although, considering the aggression you can run into at 6-max, playing this tight would be
inviting trouble. Suggestions for other positions (use entirely at own risk):
That’s another 56 hands from the SB and 78 Hands played from the Button.
Flop Math
Ok, you’ve settled on which starting hand you’re going to play, bets have been made and you
want to know what the chances are of the flop hitting your hand.
First, how many cards will help your hand improve and how many won’t or how many outs do you
have? For example, if you hold AK and you want to know what the odds are of making a pair,
consider:
When dealing the flop, there are 50 unknown cards, we remove 1 and there are 49
unknown cards, we remove 1 more and there are 48 unknown cards.
There are six flops that will give you one pair and only one pair:
Now you shouldn’t be so surprised when one of the fish who called your AK raise with Ax beats
you at showdown because he hit a pair of 5’s with his A5o and you missed your King and Ace. If
two such players call you they have a combined 65% chance to hit at least one pair.
Let’s say you have a pocket pair, what are the odds you’ll flop a set?
There are 3 ways you flop a set (assume you are holding 99):
There are only two cards that will give you a set and when 1 of them hits the flop we
assume the other one will not come out.
And there are 6 cards in the deck that can help you. Once you’ve paired your first card,
there are only 3 cards left that can pair your second card.
There are 13 cards of same suit in the deck. You hold 2 of them, that leaves 11 cards
that can help you. We need 3 of those suited cards to hit the flop.
There are 11 cards remaining in your suit. Three of those need to be chosen. Any three
cards can be chosen from the remaining 50 cards.
Essentially, the probability = how many ways can something happen / how many ways
8
anything can happen
There are 11 cards remaining in your suit, you need 2 of them to give you a four flush
draw. There are 39 cards that won’t help you (50 – 11). Any 3 cards can be chosen from
the remaining 50 cards.
8
See http://mathforum.org/library/drmath/view/61898.html
9
Calculations adapted from http://mathforum.org/library/drmath/view/56616.html
Two boards that will give you a straight; 3 cards below the two you hold, 3 cards above
the two you hold.
We need 3 consecutive cards chosen from 10 available ranks (13 ranks – 2 ranks in your
hand – Ace which can be high or low).
Note that I really don’t know if I’ve got this right and I’m still puzzling over the math for gutshots
10
and 1 gappers so I’m not even going to attempt it here .
10
See Bugs and Matthew’s calculations for straights.
For example, if you have an Open Ended Straght Draw you have 8 outs.
Essentially, the odds for drawing 1 more card are, whether it’s the turn or river card, are almost
the same. In most situations, if you have 8 or more outs you can usually draw to 1 card as the
pot generally will have at least 5 bets or will have at least 5 bets if you hit your hand.
Your pot equity is based on your share of the pot. If there are 4 people in the hand, including
yourself, your pot equity is 25% (1/4); 5 people, 20% ( 1/5); 6 people, 17% (1/6), etc. It
represents the amount you are contributing to the pot with each bet.
Your hand’s EV is determined by your outs. For example, if you have flush draw, you have 9 outs.
With two cards to come, you can expect to hit this 35% of the time. Calculated as:
If your hand equity by the river is greater than your pot share, raise. For every bet you put in you
can expect to win more than your fair share.
Of course, it’s not always that easy. If you raise from middle position and 2 or 3 players behind
you fold, you effectively reduce your pot equity. As a rule, you are better off raising for value when
you are last to act and the people ahead of you have already bet (few players will fold to one
more bet). You always need to consider how the other’s in the hand may respond.