Hacker News Reader

Comment by Raidion | Hacker News Reader

I'm a serious but hobbyist player with ~15k hands last month.

The thinking like a poker player is mostly about being upfront about your risk tolerances and then having a culture supporting people who make the best risk adjusted decision, even if it doesn't work out.

Expected Value is complicated because a 50% chance for $100 is the same as a 10% chance at $1000. It's the variance, not the EV that makes a lot of decisions hard.

You (and businesses) need to decide what risk is acceptable, communicate that clearly. Reward people who manage risk in a way that's aligned with the business even when it doesn't work out. Get rid of people who are either take too big risks, and people who don't take risks.

View on HN

I think you meant a 50% chance for $100 is the same as a 5% chance for $1000 right? Just... funny math error as I was like, wait, no... I'll take the 10% chance at $1000 every time! (unless I'm desperate for my next meal, then, as you say, the lower variance would point to higher certainty of some money vs max expected value)

EDIT: I see someone else noticed this further down than I looked in comments, sorry for dupe.

Yea messed it up, I blame pre coffee. I'm a bad poker player pre coffee too!

As someone who used to live with a pro poker player that might be the beginning but it really comes down to the psychological aspect of being able to disconnect and deal with losing a TON of money in the short term. You need to be able to quit, consistently, after losing massive amounts and then come back and do it all again the next day. Over and over. And the losing streaks can be long.

>Get rid of people who are either take too big risks, and people who don't take risks.

The alternative to getting rid of these people, is offering positions to them that better match their risk tolerances.

I also like the discourse on bankrolls and bets sizing. I think being able to correctly size your bets is such a great, albeit challenging skill to have in the real world.

https://en.wikipedia.org/wiki/Kelly_criterion may be of interest.

One major problem here is that many players dramatically overestimate their edge.

A problem for those players, an opportunity for others :)

True!

Reminds me of: https://en.m.wikipedia.org/wiki/Chinese_word_for_%22crisis%2...

One of the rules of thumb you quickly learn when studying the Kelly criterion is that overestimation can lead to ruin, underestimation can at worst lead to slower growth.

Go here for more focused practise: https://static.loop54.com/ship-investor.html

I imagine poker players don't use expected value. That's silly. You'd be all-in on a bet with positive expected value for no good reason.

They use expected returns in conjunction with size of current pot and size of current holdings.

What you say about variance is right on. It involves the payout and the odds and so on ...

They use something internally like the Kelly Criteria, probably tuned against their own model of what is the probability of each relevant outcome.

If your bankroll is small enough that Kelly vs naive EV calculations are giving you significantly different answers for what you should do, you are probably playing at stakes higher than you should be (or you're playing in a tournament, in which case the optimal strategy is indeed different than in a cash game).

mmm, you mean 5% at $1000, right?

Perhaps GP isn’t a very good poker player :)

I am not a good poker player pre coffee and apparently not good at basic math either.

Poker players use “pot odds”.

The odds of winning a hand with a future card in order to estimate the call's expected value.

Glad someone mentioned this. A 52 card deck with 3 Aces showing means there is only 1 Ace remaining (either in hand or deck).

Future cards have significant impact on odds and betting.

Pot odds are just the ratio of the current bet to size of the pot.

If you have not yet made your hand (e.g. only 4 spades) then you consider your drawing odds, which is the chance of making your hand.

But say there is a pair on the board. Then you need to consider the probability that your opponent has or will make a full house, which would beat your flush if you were to hit. Likewise the probability that someone will have a higher flush than you.

Analyzing this in aggregate is what gives you the expected value, which is what poker players consider.

If you are simply playing a cash game with unlimited buy-ins. Then assuming your overall bankroll is sufficient, you will always make plays with positive expected value, even with high variance, because it is a continuous game and you expect the total return to be positive over sufficient iterations.

If you are playing a tournament with a set number of places in the money, then you need to take into account the variance as well as the expected return because it is an episodic game and the types of plays you will make will depend on not just expected value of the single play but what place you are in currently and the number of players left.

And probably the most difficult to estimate are the implied odds. If you do make your draw, how much money can you expect to make in later betting rounds.

Implied odds are interesting in that they can make very bad hands more playable because if you do hit, then the likelihood that your opponent believes that they are winning is higher (because the chances of such a bad hand being played rather than folded immediately should be low) and thus the probability that you can elicit more money from them is higher.

Small pairs and low suited connectors are good examples of these kinds of hands. Your pre-flop chance of winning with a pair of 2's on a full table is pretty low. But if the flop is 2-K-A, you will likely get a lot of action.

Wow this is 1000x than what’s in the article - thanks.

> serious but hobbyist

> ~15k hands last month.

Is that not roughly a full-time jobs' worth of poker? (15k hands > 12k minutes? > 200 hours > 6h/day)

I play online and multiple tables. 600 hands takes ~90 minutes at 6ish tables. Each table is ~75hands/h.

I can play up to 12 at the same time, but my play suffers and I don't feel like I'm able to focus and get better.

Fair enough, thanks for indulging me!

"Expected Value is complicated because a 50% chance for $100 is the same as a 10% chance at $1000"

Is it though?

The payout is exactly the same but what op means is that you need less "50% of 100%" bets than "10% chance at $1000" to reach the same payout.

1 in 2 for the first one will pay out. and 1 in 10 for the second one will pay out.

Variance is higher for second than for first.

I think the questioner is asking if

"50% chance for $100 is the same as a 10% chance at $1000"

should be

"50% chance for $100 is the same as a 10% chance at $500"

ah yes, of course. You are correct

At first I was confused because of my mental rule of thumb that in a Bernoulli choice, p=0.5 is highest variance.

But that is when the payoff is 1, of course!

When it's not, we're really looking at payoff² * p * (1-p) and indeed the second situation has much higher variance.

"The payout is exactly the same"

Again I ask, is it really?

OP messed up their numbers—they probably meant 5% chance for $1000 ($50 EV to match .5*$100) or similar. But I think their point is still clear

Where's the best place to play online? Non-US citizen.

The golden days are sadly gone. It has pretty much all been dried up after the US crackdown. Before that, sites were full with very loosely playing Americans who viewed a loss of a $100 just 'entertainment' or equivalent of a night out. With that all gone, you have tons of very tight Chinese and Russian players these days. Lots of hunters and not a lot of moose.

Non-US? PokerStars, by a country mile. Always has been.

Reply on news.ycombinator.com