Ed Miller’s Poker’s 1%
- By Ed Miller
- March 20, 2014
- Comments Off on Ed Miller’s Poker’s 1%
The one big idea so far (the idea that guides elite players) is that poker is a game about frequencies. What matters is not if you can play pocket jacks well or if you can figure out where you’re at in hands. What matters is that you consistently, at nearly every betting decision, present your opponent with the correct frequencies of checking and betting, or folding, calling and raising. If your frequencies are close to correct, then when your opponents play with incorrect frequencies they will effectively beat themselves against your proper actions.
This is a revolution in thinking from traditional poker wisdom. First of all, to implement these ideas, no hand reading whatsoever is required—at least no hand reading in the heat of the battle. If your frequency-based thinking is advanced enough, you can go back to a “zero level” thinking where you just play your hand and ignore what your opponent may have.
Of course, very sophisticated consideration of what opponents might have has gone into the creation of the frequency-based strategy in the first place. But once the strategy is set, it’s set-it-and-forget it. You can play it and basically not even consider what your opponents are doing.
If this idea rubs you the wrong way, think about it in these two ways. First, all I’m saying is that you could program a computer to play no-limit hold’em without building in an explicit hand-reading model. You could just give it a set of immutable frequency-based instructions, let it loose against the best players in the world, and it would do just fine. In fact, as of this writing, a group of artificial intelligence researchers has claimed to have done exactly this.
Second, have you heard about online players who play twenty-plus tables simultaneously and win at a high rate? If you haven’t, these players exist.
The only way they could possibly exist is if they relied far more on this frequency-based approach than on making constant reads. No human brain could possibly make and keep accurate reads on 20, 30, or even more games at once. At least not while simultaneously playing at a high level. What these players have done is train their brains to behave much like a computer program.
This fact is also why I’ve always thought people generally overrate the advantage that a heads-up display (HUD) offers an online professional player. If you play the game at a truly high level, the information a HUD provides becomes less and less important. Many top, top players that I know—while they do often go through the motions of setting up a HUD while they play—concede that they don’t use the thing for most decisions.
“So,” you may ask, “if hand reading is useless, Ed, why did you write a whole book about it?” It’s not useless. Not at all. It’s very difficult for a human to play this immutable, frequency-based strategy I said you could load onto a computer. Hand reading—and making reads in general—is a shortcut we humans use to try to get to the correct answer without actually knowing the perfect solution. If we concede that we will never play a perfect strategy, we can fill in the gaps by making intelligent reads.
The bottom line is, however, that everything else you know about poker is secondary to the big idea in this book. The most important thing you can do is to make sure your frequencies are correct in as many situations as possible. Do that, and you will be nearly impossible to beat. Your opponents will beat themselves against you with their flawed play.
And now for a practical question. I hope I’ve convinced you by now that a frequency-based approach is very powerful. But how do you learn this frequency-based approach?
The best way I know is to explore it one situation at a time. You play for a while and record as many significant hands as you can. Then you go through each hand and determine your frequencies at every decision point by writing out hand ranges. If your frequencies look flawed, you find a solution and build ranges that conform to the correct frequencies. Then you do it for the next hand. And the next one. You do it over and over again until you build an intuition for it.
Let’s try it out here with a simple situation so you see what I mean.
It’s a $2-$5 game with $500 stacks. A player has opened for $15 from three off the button. You’re on the button, and you call with KdJd. The blinds fold. There’s $37 in the pot with $485 behind.
The flop comes Td7s2c. Your opponent bets $30, and you call. There’s $97 in the pot with $455 behind.
The turn is the As. Your opponent bets $65, and you fold.
Did you play the hand well? It’s hard to say. One could argue for folding, since you have a weak draw. One could argue for calling, since you have a draw to the nuts with money behind. One could argue for raising, since bluffing is always on the table.
Let’s develop a frequency-based strategy for all the hands you can have and see what we might do with KdJd at each point in the hand.
Preflop, you call the raise. This is certainly not a call you’d make 70 percent of the time. Since you are one of five players the opponent raised into, you five share the responsibility to call roughly 70 percent of the time. Realistically, to call a raise from a player three off the button, you need a fairly good hand to be on a level playing field.
Let’s develop a frequency-based strategy for all the hands you can have and see what we might do with KdJd at each point in the hand.
Preflop, you call the raise. This is certainly not a call you’d make 70 percent of the time. Since you are one of five players the opponent raised into, you five share the responsibility to call roughly 70 percent of the time. Realistically, to call a raise from a player three off the button, you need a fairly good hand to be on a level playing field.
Let’s assume that you’d tend to reraise AA-QQ and AK. (In actually, I tend to flat some combinations of QQ and AK in this exact scenario.) And to go with these value reraises, you’d tend to reraise some hands as bluffs. I personally choose hands such as A5s-A4s, 76s-54s, AJo, and KQo, so let’s assume you have chosen these hands to reraise with.
Your preflop calling range might look something like this:
JJ-22
AQs-A6s, KQs-KTs, QJs-87s, QTs-64s, Q9s
AQo
This range represents 168 hand combinations or 12.7 percent of all hands.
The flop comes Td7s2c.
This is a relatively dry flop in terms of flush and straight draw possibilities. But with the low high card, it’s reasonably likely an overcard will hit on either the turn or river. Dynamic boards tend to favor the player with position, and this board is either average or slightly more dynamic than average. It’s a flop that the player with position should defend with average or above average frequency.
The preflop raiser bets nearly pot on the flop. Let’s say we want to defend (either call or raise) 70 percent of our hands. That’s 118 combinations of the original 168.
Let’s start off with the shoo-in combinations to defend. We have
JJ-TT, 77, 22
ATs, A7s, KTs, JTs-87s, QTs, 97s, 75s
That’s sets, top pairs, and middle pairs. I excluded the pocket pairs 99 and 88 from the shoo-in list because I’d generally prefer to defend 97s middle pair versus 88 because of the extra outs when behind.
This is 46 combinations. If we defended only these shoo-ins and folded everything else, our folding frequency would be 72 percent—way, way too high. We need 72 more combos to defend to hit our target.
The next obvious hands are 99 and 88, so let’s throw those in. And then I’d go with big aces and gutshots. These categories add the following hands:
99-88
AKs-AJs, J9s, 86s
AQo
That’s 44 more combos. It’s still not enough to reach our goal of 72. We need 28 more. Next I’d go to weaker aces and overcards that include a backdoor flush draw. This category includes the following hands:
A9s, A8s, A6s, KQs-KJs, QJs – all combos of these hands except those suited in hearts
That’s 18 more combos. You need just 10 more. Take your pick from hands such as Q9s with backdoor draws, 66, 55, and overcard hands without backdoor flush draws. These hands are all pretty weak, but some hand has to be the weakest hand you defend. And the bar for defending is fairly low—it’s zero dollars of profit on the $30 call (including the $37 in the pot). In other words, the weakest hands you defend should be pretty big underdogs to end up winning the pot, since the borderline hands aren’t supposed to show any significant profit.
Let’s go with a final range that looks like this:
JJ-66, 22
AQs-ATs, A7s, KTs, QJs-87s, QTs-75s
A9s-A8s, A6s, KQs-KJs, Q9s – all combos of these hands except those suited in hearts
AQo
That’s 118 combos, exactly 70 percent of the ones we started with. For the sake of argument, let’s assume that we raise none of these combos and call with all of them. (On this particular board type, there isn’t a lot of incentive to raise, so this simplifying assumption isn’t too bad.)
You may be thinking, “Wow, Ed, that’s a lot of hands. I’d usually be folding a good chunk of these.”
Guess what. You fold too much! By folding so much, you reward your opponents for making excessive continuation bets out of position on flops like these.
“But, Ed, isn’t that how the fish play?” you ask, perhaps not yet convinced in the error of your over-folding. Sort of, yes. This is how the fish play. But this is the part the fish get right. Remember back to my pyramid example from before. The pyramid with the open top represents the call-it-all-down fish. The problem with their strategy wasn’t the calling frequencies after the flop. It was the fact that they started out preflop with way too many hands. So even though they are calling correctly with many of these hands after the flop, they have a lot of extra junk hands in their ranges that we don’t have in this example. It’s all this extra junk that they must get rid of (and can’t hide) that ultimately dooms them.
Remember that refusing to fold after the flop can be very frustrating, especially when the player refusing to fold is the one who has position and the board could change a lot on future cards. That annoying player refusing to fold is you.
So you call with your KdJd, since it’s on the list.
The turn is the As. It’s an interesting card. Many players would jump to the conclusion that this card benefits the preflop raiser. But if you look at the range of hands you called with on the turn, you’ll see that it contains plenty of aces. There’s no reason to think that the preflop raiser has an ace more frequently than you do.
It is no worse than an average card for you. Your opponent bets $65 into $97.
He’s backed off the nearly pot-sized bet on the flop with a two-thirds pot-sized bet on the turn. Because you have position, because the turn card is at least average for you, and because you are being offered better odds, you should defend with at least 70 percent of combos again.
Let’s go with 70 percent. Since the As was used in 9 of the 118 combos you called the turn with, you’re down to 109 possible combos. You want to defend at least 70 percent of these, or 76 combos.
Let’s start again with shoo-ins.
TT, 77, 22
AKs-ATs, A7s
A9s-A8s, A6s suited diamonds or clubs
KQs-KTs, QJs-98s, QTs-T8s, 86s, Q9s – all combos of these hands suited in spades
AQo
That’s 47 shoo-in combinations of top pair or better or a backdoor spade flush draw. If you were to fold everything weaker than these hands (and this is precisely what many regular no-limit hold’em players do), you would be folding 57 percent of the time! Again, this is way, way too often. I can bet two blank cards on the turn for a huge profit against this sort of player.
We need 29 more hands to defend. Let’s start by keeping JJ and the hands with a ten in them.
JJ
KTs, JTs-T9s, QTs, T8s – all combos of these hands except those suited in spades
That’s 16 hands. We’re getting close. We need just 13 more combinations.
I’d pick the open-ended straight draw 98s (not spades) next. Now we need just 10 more combos.
Candidate hands would be the gutshots: KQs, KJs, QJs, J9s, and 86s NOT suited spades. Because we didn’t carry forward the big hands with diamonds in them, that’s 12 combos.
Other possible candidate hands would be the hands with sevens or 99 and 88. I personally prefer the gutshots since they still have big hand potential while the small pair hands are hanging on for dear life at this point.
It wouldn’t be wrong to defend a small number of extra combos as well, but for the purpose of this exercise we can stop with these.
At this point in the hand, we likely want to raise our strongest hands such as TT, 77, 22, ATs, and A7s. This raise denies a cheap river card if our opponent holds backdoor spades or one of the now-numerous gutshots. It also extracts value those times our opponent holds a big ace.
If we raise exactly the hands I listed above, we will be raising 14 combos. Due to a frequency rule I haven’t yet introduced, if we raise 14 combos for value on this turn, we will also want to raise approximately 14 hands as bluffs.
The gutshots and 87s work well for this purpose since they have little showdown value, while they also have the potential to win stacks if the turn bluff is called. Backdoor flush draws with no pair are also candidates to raise. There are more than 14 combos of these drawing hands with weak showdown value, so you can pick 14 of your choice and raise them.
There is also merit in choosing one or two combos of sets and flat calling with these to preserve nutted hands in your range on blank river cards. This is a more advanced concept and it doesn’t really change things too much.
In the actual hand, you held KdJd and folded to the turn bet. Having gone through our frequency analysis, it turns out that KdJd is a marginal hand in your range. It’s right at or near the bottom of the set of hands that you might want to defend against this turn bet.
Because it’s a marginal hand, it’s likely that raising, calling, and folding this particular hand all have roughly similar EV. Because it’s clearly a superior hand to the small gutshot hands J9s and 86s, I’m likely not folding KdJd to the turn bet. I’d probably end up raising it, along with 98s, and a few combinations of low showdown value flush draws. These are the bluff combinations that complement the value raises I’d make with sets and aces up.
Back to the original question. Holding KdJd, you call the Td7s2c flop and fold the As turn. Did you play the hand well? I believe that folding KdJd is a small mistake, and raising would be a preferable play. Calling is also a reasonable option, but likely not quite as good as raising.
But if you chose to fold KdJd (and similar high card gutshot hands), and instead chose to call with 99 instead, it would make your overall strategy only slightly weaker. Because your frequencies are still solid—you are still defending the correct percentage of hands overall—a flaw in your choice of which particular hands to defend is of secondary importance.
The way you eviscerate your strategy is if you get the folding frequency grossly wrong. Start folding 50 or 60 percent of hands instead of 30, and you’re just killing yourself. If you play like most regular no-limit players, you make precisely this mistake on many turn and river cards.
The range-building, frequency-based analysis I just did on this hand, this is what you do to learn to play like an elite player. You do this analysis over and over again. Every time you play a session, you write down at least 1 to 3 hands you played and then perform this sort of analysis.
The hands you choose need not be the biggest pots or the nastiest beats. My example, for instance, is a seemingly mundane hand. You flopped overcards, took one off, and folded your gutshot to a turn bet. A hand like this one is unlikely to be one that you think about for days afterwards. But these decisions are the bread-and-butter of no-limit hold’em. These situations are where the edges are made and lost. It’s all in the frequencies. You want to make sure your frequencies are solid in hands like these. Make sure your pyramids are smooth. Don’t give your opponent opportunities to bet two blank cards and beat you.
We’ll go through more examples like these as the book proceeds and I teach you a few other key concepts. But I wanted to address the complaint I hear most frequently when I introduce students to this sort of analysis. “How am I going to do all this at the table? I’ve got ten seconds to make a decision. How the heck am I going to count through a hundred hand combinations to get my frequencies perfect?”
You aren’t. This analysis is not done at the table. The only thing that happens in those ten seconds at the table is that your brain recognizes similarities between the present hand and ones you’ve seen before. Then you make an intuitive decision. Your brain—after repeated analyses away from the table—learns to program itself to make these decisions much like a computer.
It’s muscle memory, except no muscles are involved. It’s the same way the violinist knows in a split second exactly where to put her finger to play the note she wants. It’s the same way the tennis player knows precisely how hard and at what trajectory to swing the racket so it ricochets the ball hard and straight down the line. It’s muscle memory. You train your brain in practice, and then in the heat of the moment it gives you instant feedback. The more you train, the more sharply accurate the message from your brain.
So learning to play poker like the 1% is both simple and hard. All you have to do is analyze the hands you play in the manner I analyzed the example above. The hard part, of course, is that you must do it thousands of times on all different sorts of hands to become elite.
Did you like this? Buy Poker’s 1% here.
Latest posts by Ed Miller (see all)
- Ten Skills To $5K - June 16, 2014
- Ed Miller’s Poker’s 1% - March 20, 2014
- Don’t Play No-Limit Hold’em Like It’s A Slot Machine - December 20, 2013
