Our Recent Essays Behind the Front Page
Tuesday, January 31. 2017
My poker night is this Friday, and I look forward to it each month. 50 people, food, libations, conversation and competition. There are afew heated moments. Not many, mostly good fun and the chance to take your friends' and neighbors' money. I have more or less broken even in this game. I keep pretty good records. We play a cash game for an hour, then a tournament, followed by cash games as people are bounced.
It's a game of incomplete information. Computers are good at games that have lots of available information, not games lacking in it. But poker often has more information available than we care to admit. Betting trends are like prices. They give signals. Being able to sort through the signals and recognize where value lies is a skill. But humans also think and behave in linear ways. If you have a non-random betting approach, a computer should be able to sniff it out.
A computer beating a human at Texas Hold 'em, even a limited table such as this, is quite a feat. The fact I happen to know the inventor of the computer makes it doubly interesting for me. He presented some of his yield optimization techniques to our team several times. I have suggested to our team we bring him back in.
Display comments as (Linear | Threaded)
this works in a multi-way no-limit game? its got to do more than brute-force odds calculation, its got to be able to read several personalities, as implied odds can be manipulated.
there's a huge difference between low limit (no-foldem holdem) and high limit play as well, but the betting units are just chips, how can a computer assign a meaningful value to a chip that might mean a quarter ($.25) or a Quarter ($25)?
I used to participate in, and regularly hosted, a monthly, sometimes weekly poker game. I think you may have inspired me to try, 15 years later and 40 miles farther, to organize a new game.
Thank you in advance,
(Mid Coast Maine, if any Maggie's readers wish to join!)
Go, Checkers and Chess have extremely specific rules, not really complex at all and hence subject to algorithmic analysis.
Watson's win at Jeopardy was impressive, but in truth Watson did no understand a single question/answer in the way a human does. It basically searches through pattern matching until it finds one that has a sufficiently high score. A human parses the question, but then uses that information to build a mental model of what is being searched for. This is not an algorithmic process.
I was thinking the other day of the gag in one of the Pink Panther movies
Does your dog bite?
Nearby dog bites him.
I thought your dog doesn't bite.
That's not my dog.
We see humor in this specifically because the human does not literally just parse the question. He constructs a model, which includes a rational deduction of why someone would ask about a dog biting. And putting himself in the other speaker's mind, realizes that person has misinterpreted the scene. No one teaches a person to do that, a computer would need to have that scenario embedded in it's behavior pattern. It does not understand why that particular question was asked, it does not have the ability to put itself into the thought pattern of the individual asking.
I read about a computer that was not quite as successful as this one, which ran in a tournament last year. Basically it played however many billions of potential hands could be played out, and scored each one, then used the odds to enable its betting. It was incapable of accounting for the information the other players were providing (or not providing) and simply 'played its hand'.
It did well enough, but nowhere near as well as this one.
Apparently, the inventor used the information provided in other bets, as well as information available about the odds of other potential hands that could be held, and the betting patterns used by other players, to account for the decision making.
It's a close approximation to what I do when I sit at the table (though I have a harder time calculating the odds).
Basically, I will watch betting behaviors to determine whether or not I think someone has a better hand or not. I'm willing to fold to a very strong bluff if I don't have a moderately good hand.
It's a strategy that pays off. But you can get suckered into a slow play of a stronger hand.
I'm pretty impressed by whatever process he's using. Poker is not an easy game. Luck plays a role, but it's definitely not the only factor....not even a major one, if you bet properly.
I don't get the concept of luck in poker. there's no luck, only probabilities, and someone could draw to inside straights with the correct implied pot odds with a positive expectation.
and I'd be careful about judging based on betting patterns. a preflop checkraise followed by the check-raiser betting on the flop is standard in a serious game regardless of whether the flop hit and could mean anything from a bluff to a made hand.
You're talking about a serious game - I've rarely played with people that good. Even so, that's not a pattern I'm referring to.
I've noticed very few - even the best - players randomize their behaviors. Recently, an amateur won the WSOP and he won with an incredible number of bluffs. That was about as random a pattern I've seen in a long, long time.
And I also get a bit teed off when people talk about 'luck' in poker. But there is an element.
let's say I've drawn 3 of a kind and my opponent has a low gutshot straight, at the river. I bet heavily, trying to drive him out. An experienced player will likely fold, unless he's short stacked and his back is against the wall. Assuming he's even moderately experienced and has a decent stack, I'm walking away with the pot. But I've seen plenty of people hang on, match the bet, and hit a highly improbable gutshot on the river.
That's the luck. There's no question it plays a role and if you have the cojones or desire to play on luck alone, you can grab a few big pots here and there. But you're more likely to lose bigger pots along the way.
If I'm holding a gutshot straight and, all other things being equal, I'll call if the pot presents a payback of 9-1 or more. the thing is, the money favorite bet and the odds favorite call are both correct.
obviously the trips will win most of the time, but the made straight will pay for all the losses.
If you've seen 9-1 payouts that often on hands like that, I give you credit for playing in higher stakes than me.
Regardless, as you point out, the trips are likely to hold, and luck is still a factor even if you calculated them well enough.
My point is I've seen betters do it for far less, and the issue on the river is still luck. You can say it's just the odds, but ultimately it becomes whether or not you're lucky enough to have the odds play out your way. They are still very low probability.
Also, I agree the one win will pay for the other losses in a cash game. Given the nature of tournament play, that is less assured.
but that's the point. the reason both bet and call are good actions is because over the long haul, betting on an inside straight that gives at least 9-1 return (on 8-1 odds) still shows a profit even if the straight stays broken most of the time. remember, this isn't heads-up play.
I'm not going to get those pot odds (9-1) unless there are other players calling your bets. 8-1 are the odds of making the straight, so I'd be careful with small straights (don't want to be caught with the ignorant end) and, depending on the situation might not call anyway. so aggressive play with your trips is the right tactic, thinning the herd to give drawing hands less than the correct odds to call.
the most alarming thing to hear in poker: "how big is your full house?"
true story: common in Las Vegas are bad beat jackpots. a small amount is raked from each pot and put aside into a bad beat jackpot. meaning, whoever loses with (depending on the house) at least quads or maybe aces full, gets one half of the jackpot, the rest is shared with the table. at the Excalibur the jackpot had reached somewhere north of $20,000 when, at a 7 card stud table, a queen high straight flush was losing to a king high straight flush and the table was beginning to celebrate. but it was only 6th street, and the river card was a king suited to the queen high straight flush. no bad beat, chopped pot. the guy who lost the bad beat came flying out of his seat at the dealer, hands around his throat, choking him until the floormen dragged him away.
I understand your point. I think you're missing mine.
Whether you play the odds properly or not, there is still an element of "luck". Sure, your example has a positive outcome over the broad spectrum of that type of bet. Lots of iterations required.
However, that's a cash game. Tournament odds (since the cash you potentially lose each time can keep you out of the money) would be slightly different. If you're going to go bust, do you still play the 1-9 odds for the 9-1 payout? Big question - tough call.
At any rate, I'm merely pointing out that you're assigning professional guidelines to a situation where MOST PEOPLE don't know how to figure it out. As a result, I've been beaten by betters who have no business being in the game at that stage, simply because they hung around to see the river, even though the payout for them wasn't that good. It's luck. That's it. They are playing for luck.
Love the bad beat story. I've never been at a bad beat table which won, but I know people who have been. Friend of mine lost out 4 Kings to straight flush. Out of the tournament, but bad beats applied, so he came out (as did everyone at the table) well ahead for the tournament. Not a terrible way to lose.
the structure of the rising blinds and go-for-broke mindset makes these games different, but here's a solution you can try in any of the free online tournaments. buy into a game and then just watch for about 20 minutes until the maniacs have had their way and most of the players are gone. you'll be down your blinds but otherwise in good shape (because its still your stack vs every other chip that isn't yours, just as it was when the tournament began). now you can play aggressively with knowledge of tactics and odds for loose shorthanded games and you should come out alright because they are maniacs and you are not.
true story: my gun club held poker tournaments after league nights, and I had the smallest of the three remaining stacks, small enough so that I knew I had no chance of winning, but the payout included the second place finisher. I decided to fold every hand, not defend any blind, assuming correctly that in the four or five hands I had left before I was blinded out, the big stack would put the smaller stack all-in and if he won, I'd finish second. which is what happened, with me holding a single chip and the second place prize money.
Or, as Branch Rickey once said "Luck is the residue of good planning."
Another saying is "Luck favors the prepared mind."
In either case, you're basically saying luck is just probability. Nobody would dispute that. But when something is 89% unlikely and it happens, and you just happened to call it...well, most people will call that luck. You can say "I had a prepared mind, or I had good planning." Yes to both, but you were still betting against the odds.
I'm looking at it differently. if I ask you whether a 100-1 bet is good or not, you can't answer the question without knowing the payout. if the payout is 101-1, the bet is good because it shows a net positive expectation.
I'm not thinking of a poker game as something that lasts an evening. I'm thinking of all the hands I've played and will play. that's, what, tens of thousands? hundreds of thousands counting casual online play? with that sample size, I'll happily make those bets, knowing I'll lose almost all, but also knowing the probabilities will be at or vanishingly close to the true odds, which makes those bets long term winners.
so if you play high suited cards preflop, you are a likely winner if the flush fills, which it will about 20% of the time, and a likely winner if the high cards pair even if the flush doesn't fill. AK is a small favorite over an aggressive any JJ or lesser pair, so if the flush odds don't materialize, you've got a reason to keep with the draw and another way of winning. here you can bet in ways to disguise the flush draw as a high pair or a pair as a flush draw, and, a bluff on the river might win the hand anyway.
I'll accept losses when things go sideways because probabilities aren't guarantees. calling it luck is superficial and unhelpful labelling (luck = good outcome; bad luck = bad outcome). Understanding probabilities means understanding how the game (in fact, how the universe) works. I'm prepared in that I have a feel for the other players and a grasp of basic math. but luck describes an outcome, but doesn't tell you how to get there.
Thought Kriminal No. 392342-A: I don't get the concept of luck in poker. there's no luck, only probabilities
The fewer the hands played, the greater the influence of luck. All card players have experienced a night where bad or good cards predominated.
Luck has a normal distribution, and you won't always be at the mean. That's probability!
you'd be welcome in any game I'd play in. we'll give you an ice cold beer, make you feel at home, be your best friends for the evening, and when you tap out, we'd invite you back. rinse, repeat.
One night, in 1965 as I recall, I sat down at a 5 card draw table in Gardena, CA. Limit game - $1, $2, that's like $10, $20 today. I had an extraordinary run of cards. In an hour I was dealt about 10 pat hands. Problem was that nobody had as much as openers on every one of them. It's very frustrating to win the ante with 4 kings.
Last month, at my tournament, I won the first 5 hands, all of the all ins. I started out the night up 35,000 chips.
The extraordinary run took a turn on the 6th hand when my 6th consecutive all in (2 pair K's and Q's, which I had on the flop) was beaten by three 3's on the river.
By the time the break rolled around, I was down to 18,000 chips....mainly because you can't sit by when you get dealt Q 10 suited, K 10 suited, 9's, etc. I don't get crazy aggressive in my bidding either.
I don't like winning big early, especially if you keep getting great cards after you win big early. I've made the final table more often on nights when I get hammered early, have to start playing conservatively, and then pounce on that one great hand that rolls around.
I feel for you, though. It's part of the reason I've lost interest in standard games of poker and enjoy Hold 'em more. There are lots of non playable hands in 5 card draw that people suddenly feel compelled to play in Hold 'em.