Do you believe in Luck?

[Antiquehunter]

Does your husband only think he's "unlucky" on these particular coin-toss hands?

That is my perception here, yes. Because I have offered many other examples of potential 'unlucky tests' we could do.

- Dealt a pair and flops a set more or less often than predicted mathematically.

- Gets all-in with one other player with other scenarios that are either close to a coin-toss or have an otherwise easily established expected win frequency.

We keep coming back to the all-in with the pair vs two overs. No meaningful data has been provided to show overall results (a few million points in 'free' chips, and an unsubstantiated track record in tournaments & live games).

I personally track EVERY game I play in a spreadsheet, so that I can derive an expected hourly return. I separate this data recording by tournament & cash game data.

It is a dangerous (and losing) attitude to have a feeling of being 'unlucky' even at one isolated type of card playing event. I'm interested in seeing this resolved to improve this guy's game, not to try and prove/disprove any actual sense of someone being 'unlucky'.

 

[Beth]

Beth - if he doesn't want to gather more data online, but really is curious about testing his 'luck'...

Just get a deck of cards. Deal out a QQ and an offsuit AK where the suits differ from the QQ. Have him choose to be the 'favorite' (QQ) or the 'dog' (AK) on either side of the 52/48 proposition.

I don't think this will be of help because it doesn't test the theory in a game situation. Thanks anyway.

Again, I think what's most likely is that he's misperceiving even money gambles. (That is, the odds really aren't 1:2.)

And it could also simply be confirmation bias, something your efforts would clear up.

This was my suspicion and what I expect this effort will eventually show.

Does your husband only think he's "unlucky" on these particular coin-toss hands?

No, he thinks he's unlucky on all showdowns. He picked these hands because they are 50/50 and at first he just recorded whether he won or lost.

 

[JoeTheJuggler]

From this page, the minimum sample size where this difference would be significant would be 129 for a two sided (testing the hypothesis that either he's lucky or unlucky, but luck is a factor) or 105 for a single sided test.

From this site:
http://www.quantitativeskills.com/s...ethod=OneSample&ConCor=true&Submit1=Calculate

From the same site, the result of the t-test is not significant at 5% level:
http://www.quantitativeskills.com/s...=00.00&CI=95&onesample=true&Submit1=Calculate

(Though it gives a p of 0.94871, so I assume that means it would be significant at the 10% level. But again, you should establish the confidence level as part of the hypothesis. Otherwise, you can always claim that it's significant at *some* confidence level if the result is anything other than exactly 27 out of 54. ETA: And if you use weasel words like "not significant but tantalizing" or the ever-popular "nearly significant", then you pretty much can prove any hypothesis you want at least merits more study.)

 

[Antiquehunter]

I don't think this will be of help because it doesn't test the theory in a game situation. Thanks anyway.

Beth - how does what I describe differ from a game situation?

If he's saying he's unlucky with this PRECISE situation (all-in pre-flop pair vs two overs) then what difference does it make game vs non-game?

Just because there are chips at stake doesn't change the fact that its a 52/48 (roughly) prop. Now - YES it is true that his decisionmaking process leading up to the point where the all-in is made & called (either called by the opponent or called by him) may be flawed. But at the point that we are down to two players all-in, there is simply no more skill involved. Whether or not it was the most prudent thing to do, if he has QQ and went all-in and got called by AK, he is now a 52/48 favorite. That's it, point blank.

So testing QQ vs AK in a live or a computer simulated environment is exactly what would show your husband if he is indeed more lucky/unlucky than expected. I proposed dealing the cards out manually so he feels more directly 'involved' with the game - very important to shuffle appropriately for obvious reasons.

If he feels that the game requirement is necessary to adequately test his 'luckiness' as has been described to this point, then I think we're getting into something much more than mathematics - we're getting into the realm of woo.

 

[JoeTheJuggler]

No, he thinks he's unlucky on all showdowns. He picked these hands because they are 50/50 and at first he just recorded whether he won or lost.

First, then Fredrik's point is correct. Analyzing such a small portion of those hands wouldn't answer his question (whether or not he's unlucky on all showdowns.)

Also, since you don't get to showdowns without the application of skill (or rather decisions based on skill or lack thereof), his claim is now indistinguishable from the claim that he's not a very good player.

That is, if you looked at the two cards he was dealt for every hand where he was in to the showdown, and those cards showed a significant trend to being worse than what you would expect if they were dealt randomly, it would only mean that he should've folded hands that he didn't. And playing to showdown involves a LOT of stuff that is not at all random even if we accept the null hypothesis that there is no such thing as luck. (For example, a really good player can win pots with virtually any two cards in the right circumstances. As has been said, you gotta know when to hold 'em and know when to fold 'em, and if you play just strictly by the probabilities, you will be destroyed by good players because playing so transparently is one of the easiest poker strategies to beat.)

 

[Finsend]

Beth - if he doesn't want to gather more data online, but really is curious about testing his 'luck'...
<snipsnap>
I think we've given you all the pieces to the puzzle in this thread - either he genuinely wants to explore the hypothesis, or he's made up his mind and isn't interested in testing it.

This seems interesting to me

You could find out if his bad luck is the same in a test situation (no real stakes) compared to a real situation! If there are changes his bad luck statistics this then might have to do with signals or at least playing differently under different circumstances... I could have missed out on something -in that you already did this and the statistics are test situations or mixes real and tests...

It just interested me

If there are no differences, then you know at least he is playing the same way in both situations - so what I am saying is does it matter whether he pays against a statistical robot (for real / or not) or against a human player (who might be able to read him, etc) (for real or not) etc:

Seems he has to play a lot of extra games, maybe, geheh



cheers,
F

 

[Antiquehunter]

Without wanting to further complicate matters...

Amarillo Slim, a very successful and famous poker player & proposition gambler, will take very heavy action on the following wager:

The opponent gets to choose first from any of these three hands:
22
AKs
JTs

Slim will choose his hand separately.

Then, for say a $1000 wager, they will deal the cards out 100 times. The person who wins the most hands out of the 100 deals, gets the $2000.

If you pick 22, Slim takes JTs
if you pick AKs, Slim takes 22
If you pick JTs, Slim takes AKs

He enjoys a slight edge in each matchup, and trusts that over 100 deals, this slight edge will be the winner. I suspect he doesn't get many takers anymore, with the mathematics being readily available, but he does STILL get action on this proposition.

To put this in perspective of your husband, Beth - he should try this proposition, employing Slim's strategy. Bet at a stake that is comfortable for him. Of course, it is definitely a 'loseable' bet - as the edge is very small. But he should end up an overall winner, playing out the same bet over and over again...

Bob Stupak once said words to the effect of 'Give me a .01% house edge and a big enough bankroll, and I'll break anyone who plays in my casino'. Clearly he wasn't afflicted by your husband's (supposed) problem.

 

[JoeTheJuggler]

To put this in perspective of your husband, Beth - he should try this proposition, employing Slim's strategy.

Except, of course, you can't know what hand your opponent has when you make the wager, as in Slim's wager you described. I think it hinges on the misperception that choosing first is an advantage, when in fact it's a disadvantage.

It'd be like playing rock-scissors-paper where you allow your opponent to choose before you do!

 

[Antiquehunter]

But my point Joe, is that if we are testing only whether or not if Beth's husband wins around 52% of the time he has a pair vs two overs, then it is immaterial whether or not the decision to go all-in (or to call an all-in bet) was the 'right' one. At that point in time, once the money is in the middle, its a 52/48 prop. No more influence of 'skill' is involved.

That is why a simulation of dealing cards onto the kitchen table, or in a computer scenario is viable.

If the problem is 'I lose too often playing cards, and I think its because I'm unlucky on heads-up showdowns of pair vs two overs' - then this is an entirely different kettle of fish.

 

[JoeTheJuggler]

And again, I implore people to try to use the word "luck" consistently rather than ambiguously in this thread.

I'm pretty sure Beth has agreed that the hypothesis she is testing is whether or not the outcomes are explicable by her husband's bad luck and the null hypothesis is that only random chance and not luck explain the outcomes.

"Luck" as something it makes sense to believe in or something that "happens" or can be an explanation for anything can only be defined as in Merriam-Webster's first definition which I quoted earlier.

To use it in the purely "descriptive" way (using Skeptic Ginger's terminology) is confusing. To say that one is "lucky" any time he wins something independent of skill is using the term a different way. It's not offered as predictive or as a cause or explanation of anything--or something one can "believe in". (In fact, this usage is pretty meaningless if you think about it. At most it's only the observation that an outcome was undeserved.)

 

[Antiquehunter]

Well, I'll put my cards on the table...

'Luck' as I use the term in poker / gambling, is where I manage to win an event, when I am an underdog. I am 'lucky' when I win a 48/52 poker situation. I am 'lucky' when I happen to flop the nut straight to my 87. I am 'lucky' when I place the 6 at craps, and the next roll is a 6. I am 'unlucky' when I have AA, get called by someone holding KK and he hits his 3rd K (and I don't get my 3rd A).

I don't believe 'luck' exists as an entity. One cannot be consistently 'lucky' or 'unlucky' at gaming (or at life). No one is inherently 'lucky' or 'unlucky'. Otherwise the mathematics of the game(s) would be immaterial.

On the basis of this understanding of 'luck' I encourage Beth & her poker playing husband, examine their dilemma as I propose.

 

[JoeTheJuggler]

But my point Joe, is that if we are testing only whether or not if Beth's husband wins around 52% of the time he has a pair vs two overs, then it is immaterial whether or not the decision to go all-in (or to call an all-in bet) was the 'right' one. At that point in time, once the money is in the middle, its a 52/48 prop. No more influence of 'skill' is involved.

That is why a simulation of dealing cards onto the kitchen table, or in a computer scenario is viable.

If the problem is 'I lose too often playing cards, and I think its because I'm unlucky on heads-up showdowns of pair vs two overs' - then this is an entirely different kettle of fish.

I agree that your simulated game would test the hypothesis if that hypothesis was only limited to these kinds of hands. I think the dismissal that not being in a real-game situation is unwarranted since the "unlucky" hypothesis explicitly depends on the premise that everything else about the real-game situation is eliminated.

But it's like any number of paranormal claims (like those brought to JREF's MDC): it's up to the claimant to clarify what the claim is, and they often don't make any logical sense. (Such as, "I can see the future, but I can't see tomorrow's lottery numbers," or variations on that theme.)

So, I have no problem with testing it from just actual hands played in the game. I would caution against the pitfalls of a lot of paranormal testing though: no arbitrary stopping and starting (as Beth was already doing by trying to find significance in a subset of the 54 trials) and some way to control the data collection, given that the one person collecting that data is biased. (There are lots of tests that show that the bias of the person collecting the data can result in inaccurate data--either by "sheep" or "goats".)

And absolutely no allowing the discounting of a trial for any reason at all. (As in, "That one doesn't count because _____" and fill in the blank with anything at all, not even stuff like, "I left the room before the remaining cards were dealt so my unlucky juju wasn't there.")

 

[JoeTheJuggler]

And here you're using both of these two meanings.

First, the one Skeptic Ginger referred to as merely descriptive:

'Luck' as I use the term in poker / gambling, is where I manage to win an event, when I am an underdog. I am 'lucky' when I win a 48/52 poker situation. I am 'lucky' when I happen to flop the nut straight to my 87. I am 'lucky' when I place the 6 at craps, and the next roll is a 6. I am 'unlucky' when I have AA, get called by someone holding KK and he hits his 3rd K (and I don't get my 3rd A).

And then the one I think Beth's poll is about (the one in the first M-W definition):

I don't believe 'luck' exists as an entity.

If by entity you mean something that can cause or explain outcomes.

The trouble is here, where you conflate the two:

One cannot be consistently 'lucky' or 'unlucky' at gaming (or at life). No one is inherently 'lucky' or 'unlucky'.

And I think Beth is guilty of this conflation also in saying that if the results of further testing get closer to the expected values that it means her husband's "luck" is evening out. If you take that approach, then the "Yes" answer to the question, "Do you believe in luck?" is supported no matter what outcome you get. (Like in the usage you first used.)

I can see the kind of connection you're trying to make, though, between the two meanings. You could define the predictive/explanatory/causative type of "luck" as being the idea that some people are prone to being lucky or unlucky using the purely descriptive type. But this is just confusing things. [ETA: It changes the purely descriptive into something predictive as if it's the same meaning.]

Either we're testing a hypothesis about a predictive/causative thing that can be explanatory or we're not.

If we're not, as I said, there is no outcome that can help one rationally answer the question, "Do you believe in luck?"

 

[Finsend]

And again, I implore people to try to use the word "luck" consistently rather than ambiguously in this thread. <squoiks>At most it's only the observation that an outcome was undeserved.)

Wiki Luck or fortuity is good or bad fortune in life caused by accident or chance, and attributed by some to reasons of faith or superstition, which happens beyond a person's control.
http://en.wikipedia.org/wiki/Luck

Oxford Luck is success or failure apparently brought by chance rather than through one's own actions:
As In:
"Becasue he could not play soccer at all, without them legs and arms, it was just sheer luck that the first kick went in."
"Them thingies are supposed to bring some good luck, aren't they?"

But also as in:

- Chance considered as a force that causes good or bad things to happen (this would be Gargamel or Gandalf being involved).
- Something regarded as bringing about or portending good or bad things:
I don‘t like Friday — it’s bad luck (also pretty superstitious
http://oxforddictionaries.com/definition/luck

So for me Luck is well... just that!
It sure exists as a word and as a matter of saying.
But it is - 2me - something different than the other word ehm I forgot that one.

Maybe chance?

Oxford dictionary: a possibility of something happening:
there is a chance of winning the raffle
there is little chance of his finding a job
(chances) the probability of something desirable happening:
he played down his chances of becoming chairman
[in singular] an opportunity to do or achieve something:
I gave her a chance to answer
the occurrence of events in the absence of any obvious intention or cause:
he met his brother by chance
http://oxforddictionaries.com/definition/chance

SO luck is caused by a possibility of something happening!
yes

 

[Beth]

From this page, the minimum sample size where this difference would be significant would be 129 for a two sided (testing the hypothesis that either he's lucky or unlucky, but luck is a factor) or 105 for a single sided test.

From this site:
http://www.quantitativeskills.com/s...ethod=OneSample&ConCor=true&Submit1=Calculate

From the same site, the result of the t-test is not significant at 5% level:
http://www.quantitativeskills.com/s...=00.00&CI=95&onesample=true&Submit1=Calculate

Nice site. The situation I'm working with would be a one-sided one-sample test. Our sample size will increase over time.

ETA: And if you use weasel words like "not significant but tantalizing" or the ever-popular "nearly significant", then you pretty much can
prove any hypothesis you want at least merits more study.)

I have to disagree with the "not significant but tantalizing" and "nearly significant" descriptions. I wouldn't describe them as weasel words. I think they are reasonable descriptions of a p-value of say 0.08 or .13 but not .4 or .72. There is very little difference in a test that comes out with a p-value of 0.04999 versus one that 0.05001 and it makes sense to include such qualifiers when you are dealing with a situation that is close to the border between rejecting the null and not rejecting the null. After all, there isn't much difference between a p-value of 0.04999 and 0.05001.

Beth - how does what I describe differ from a game situation?

It's not a game situation. Sorry, please don't feel rejected by this, it has nothing to do with you or with the mathematics of it. The idea of testing via a simulation simply isn't going to work for this particular situation.

I'm pretty sure Beth has agreed that the hypothesis she is testing is whether or not the outcomes are explicable by her husband's bad luck and the null hypothesis is that only random chance and not luck explain the outcomes.

Yes. If the null is rejected, we can assume it's random chance and confirmation bias.

 

[JoeTheJuggler]

Wiki Luck or fortuity is good or bad fortune in life caused by accident or chance, and attributed by some to reasons of faith or superstition, which happens beyond a person's control.
http://en.wikipedia.org/wiki/Luck

Oxford Luck is success or failure apparently brought by chance rather than through one's own actions:
As In:
"Becasue he could not play soccer at all, without them legs and arms, it was just sheer luck that the first kick went in."
"Them thingies are supposed to bring some good luck, aren't they?"

But also as in:

- Chance considered as a force that causes good or bad things to happen (this would be Gargamel or Gandalf being involved).
- Something regarded as bringing about or portending good or bad things:
I don‘t like Friday — it’s bad luck (also pretty superstitious
http://oxforddictionaries.com/definition/luck

So for me Luck is well... just that!
It sure exists as a word and as a matter of saying.
But it is - 2me - something different than the other word ehm I forgot that one.

Maybe chance?

Oxford dictionary: a possibility of something happening:
there is a chance of winning the raffle
there is little chance of his finding a job
(chances) the probability of something desirable happening:
he played down his chances of becoming chairman
[in singular] an opportunity to do or achieve something:
I gave her a chance to answer
the occurrence of events in the absence of any obvious intention or cause:
he met his brother by chance
http://oxforddictionaries.com/definition/chance

SO luck is caused by a possibility of something happening!
yes

I appreciate that one of the meanings of "luck" is synonymous with "chance" or the idea that something isn't caused by anything (i.e. outcomes caused by random chance).

However, that meaning isn't consistent with the way the word is used in the thread title and poll or in the way Beth's husband is clearly using it.

It's offered as a predictive/causative/explanatory thing.

Again, the meaning in question, the one Beth's husband is talking about is most like the first Merriam-Webster definition.

 

[Antiquehunter]

And I think Beth is guilty of this conflation also in saying that if the results of further testing get closer to the expected values that it means her husband's "luck" is evening out. If you take that approach, then the "Yes" answer to the question, "Do you believe in luck?" is supported no matter what outcome you get. (Like in the usage you first used.)

I can see the kind of connection you're trying to make, though, between the two meanings. You could define the predictive/explanatory/causative type of "luck" as being the idea that some people are prone to being lucky or unlucky using the purely descriptive type. But this is just confusing things. [ETA: It changes the purely descriptive into something predictive as if it's the same meaning.]

Either we're testing a hypothesis about a predictive/causative thing that can be explanatory or we're not.

If we're not, as I said, there is no outcome that can help one rationally answer the question, "Do you believe in luck?"

Well first off - I've been posting in the thread, but purposely have not voted in the poll. Because of the wishy-washiness of the term 'luck'.

I don't 'believe' in luck. I do observe that things happen that are statistically improbable - whether that be winning a 48/52 proposition, or winning the lotto 6/49. However those circumstances are all explainable mathematically. For all intents and purposes, if I am the benefactor of the random occurrence that puts more $ in my pocket, I can call it 'luck' for lack of a better word. Semantics would dictate that I made the decision to participate in the first place, ie I 'created' my own luck. I'm not that fussy about it. If I participate in an event that has a 48/52 chance of occuring, I will win some of those, and I will lose some. If I participate in a 1/10,000,000 event, then I will win far fewer, but it is conceivable that I COULD win one. (I don't buy lottery tickets - so it would be beyond 'lucky' for me to win the lottery.)

So I don't agree with you that I'm jumbling up my 'lucks'. I'm calling the random, happenstance occurence that falls in my favour 'luck' - simply because that is the common usage. (And similarly, I use 'unlucky' on the flipside of the coin.)

As I said before - no individual is 'lucky' or 'unlucky'. Sure we know people who appear on the surface to win more than their fair share. We also know people who appear to always get the short end of the stick. When we see it at gambling, then there are mathematical explanations - pure and simple. No lucky/unlucky superpowers involved.

 

[JoeTheJuggler]

I have to disagree with the "not significant but tantalizing" and "nearly significant" descriptions. I wouldn't describe them as weasel words.

Do you recognize that the hypothesis actually includes the number of trials and the confidence level? Therefore discussing these other confidence intervals (and other sample sizes) is post hoc hypothesizing. It's equivalent to looking for streaks within the data and doing data analysis on that subset. You really can't say ANYTHING about the significance of those post hoc hypotheses.

It's not a game situation. Sorry, please don't feel rejected by this, it has nothing to do with you or with the mathematics of it. The idea of testing via a simulation simply isn't going to work for this particular situation.

Why? What about the game situation is important to the test? You explicitly want to eliminate considerations of stakes, betting, and anything else that might involve skill.

Do you recognize the problems I've mentioned? That when someone who is obviously biased is your only collector of data, you are likely to get bad data. This is one reason the simulation proffered would be better. (Also you could gather all the data in one sitting, so you'd control for a lot of other possibly confounding variables.)

 

[JoeTheJuggler]

Well first off - I've been posting in the thread, but purposely have not voted in the poll. Because of the wishy-washiness of the term 'luck'.

It's an ambiguous term (has more than one meaning), but we can glean which meaning is being used by the context.

I think you're still using both at the same time, and it's confusing.

And that results in these contradictory statements:

I don't 'believe' in luck.

<snip>For all intents and purposes, if I am the benefactor of the random occurrence that puts more $ in my pocket, I can call it 'luck' for lack of a better word.

<snip>I'm calling the random, happenstance occurence that falls in my favour 'luck' - simply because that is the common usage. (And similarly, I use 'unlucky' on the flipside of the coin.)

M-W is pretty accurate in putting the most common American usage of the word as it's first definition. However, it really doesn't matter which meaning is most common. We can tell from the context here which one is being discussed.

We're talking about hypothesis testing, so it clearly is being used as an explanation that is different than the null hypothesis (that the results are due to random chance).

If you use the purely descriptive meaning, then you'd say our hypothesis is "His luck is from a supernatural inherent luckiness" and our null hypothesis is "His luck is due to random chance". But then the question, "Do you believe in Luck?" wouldn't apply since a belief in this kind of "luck" is required no matter which hypothesis you accept.

 

[Beth]

Why? What about the game situation is important to the test? You explicitly want to eliminate considerations of stakes, betting, and anything else that might involve skill.

Because there isn't any disagreement that in such a situation, the expected random probabilities would occur. What we're testing is the hypothesis that when he is playing actual games, his luck in showdown situations is bad. Testing his luck in a simulation isn't going to be convincing.