How much would you pay to play this fair coin flipping game?

  • Thread starter docnet
  • Start date
  • Tags
    Game
In summary, the conversation discusses a game where the reward for getting a head is 2^n dollars, but the probability of getting a head decreases with each flip. The expected value of the game is found to be an indefinite amount, leading to the question of who would want to pay to play this game. Different approaches are discussed, including one where the expected value is found to be 40 dollars and another where it is found to be 41 dollars. Ultimately, it is concluded that this is a famous paradox in which probability theory does not accurately predict the outcome.
  • #1
docnet
Gold Member
799
486
Homework Statement
.
Relevant Equations
.
Screen Shot 2021-11-26 at 12.50.22 AM.png


I feel like there is something missing in my solutions, because the answers are coming out weird.

(a)

The reward for getting a head is ##2^n## dollars, the number of flips ##n## it takes to get the first heads.

the probability to get a head in the ##n##th flip is ##\frac{1}{2^n}##

so the expected value of the game is

$$E(X)=\frac{1}{2}\cdot 2+\frac{1}{4}\cdot 2^3+\frac{1}{8}\cdot 2^3+\frac{1}{16}\cdot 2^4+\cdots$$

$$E(X)=\sum_{i=1}^\infty 1 \Rightarrow \text{ the sum diverges}$$

Who would want to pay an indefinite amount of money to play this game...?

(b)

$$E(X)=\sum_{i=1}^{40} 1 = 40$$

By this logic, I would pay no more than 40 dollars to play this game

The probability of going more than 40 rounds is

$$P(X>40)=\sum_{i=41}^\infty\frac{1}{2^i}\approx 0$$

Yet, the expectation value drastically differs from the value in (a)

(c)

$$E(X)=\sum_{i=1}^{40} 1 + \sum_{i=1}^\infty \frac{1}{2^n} =40+1 =41$$

I'd pay 41 dollars to play this game

(d)

no one has that much monay. that's too many zeros.

It turns out this is famous problem called St. Petersburg Paradox. - it seems to be a rare example of when probability theory isn't rational for predicting the outcome, since the player is probably only going to win a modest amount of money in this game.
 
Last edited:
Physics news on Phys.org
  • #2
I think all of your answers are correct here.
 
  • Like
Likes Tazerfish and docnet
  • #3
docnet said:
Who would want to pay an indefinite amount of money to play this game...?
People asking questions like this expecting students to base their answer on expectation value only are doing students a disfavor. The value function for someone playing the game is generally not going to be linear* and even if it were, the variance plays a large psychological role for people’s willingness to gamble.

* For example, once you have more money than you could possibly spend, there is no added value in earning more.
 
  • Like
Likes DaveE, Tazerfish and docnet
  • #4
Orodruin said:
* For example, once you have more money than you could possibly spend, there is no added value in earning more.
Tell that to Elon Musk, Jeff Bezos etc.
 
  • #5
PeroK said:
Tell that to Elon Musk, Jeff Bezos etc.
Even so, I do not believe for a second their value functions are linear. Logarithmic perhaps
 
  • Like
Likes Tazerfish
  • #6
PeroK said:
Tell that to Elon Musk, Jeff Bezos etc.
Fair, some people are keen on accumulating for the hell of it or for power, fame or their ego.
Some people want to be the winner of capitalism, to control armies of works if not armies of soldiers.
Although the value of more money never drops to zero, it gets real close.

Going from living on the street to 40k a year (stable housing, secure water, food and healthcare) is probably much more valuable than going from 40k to 1m a year unless you really get off on the power.

Also, for different theories of personal value/decision making, you can have a look at Daniel Kahnemannn'sProspect Theory.
 
  • #7
This question may be much more about human psychology than probability. We are simply irrational in dealing with low probability, high reward (or cost) choices.
 
  • #8
I think to answer the last question, more justification should be given than just "that's too many zeros."

Say "that's about a trillion dollars," and note this is on the order of the largest corporations' valuation. But no individual has that much money.

Something about the game though - why play at all if the expected return is just to break even?
 
Last edited:
  • #9
Everyone has missed the point, the question is not "what is the expected value of the payout from this game", it is "how much would you be willing to pay to play this game".

If you have three different answers for the three different scenarios then you have not thought about it clearly enough. Why should an event with a probability of 1 in 240 make any difference to anything?
 
  • #10
pbuk said:
Everyone has missed the point, the question is not "what is the expected value of the payout from this game", it is "how much would you be willing to pay to play this game".
I certainly did not interpret the question as "what is the expected value of the payout". Quite the contrary. I explicitly talked about non-linear value functions, which are personal and subjective and a direct representation of when a person decides they are willing to play the game. So no, I do not subscribe to this description of this thread.
 
  • #11
There are two aspects to the problem if we are thinking about it as a serious proposition. The first is that whoever is offering the game must have limited funds. This could be established. If not, then you'd need some prior credence of the maximum payout.

Second, it is a personal (not irrational) decision about the diminishing value of winning beyond a certain level. It's not necessarily the case that winning $20 million is significantly better than winning $10 million. You could, I guess, transform a monetary amount into a diminishing happiness value. And that would give you a rational approach.

Related to this is the same calculation for the negative effects of losing money. Most people could afford $5, but $1000 would hurt most people enough to deter them from gambling it away. You'd need to be lucky not to lose close to your original stake in that case.
 
  • #12
PeroK said:
There are two aspects to the problem if we are thinking about it as a serious proposition.
I think there is another, much more important, aspect: if you pay any more than $16 you have a more than 90% chance of losing money.

And another, slightly less important, aspect: if you pay what seems like a modest amount of, say $35 you have to play the game 45 times to have a better than 50% chance of winning even once, and the number of games you have to play to have a 50% chance of coming out on top overall is ... a lengthy calculation.

To pay any amount more than $2 to play this game you've gotta ask yourself a question: "Do I feel lucky?" Well, do ya, punk?
 
  • #13
I guess I'm being dense today but how does "how much you're willing to pay" come into the game? Is it 1 buck for each coin toss? If I bet 20 bucks and heads comes on toss 3, do I get 2 cubed (=8) plus my 17 (20-3) bucks back?
 
  • #14
gmax137 said:
I guess I'm being dense today but how does "how much you're willing to pay" come into the game? Is it 1 buck for each coin toss? If I bet 20 bucks and heads comes on toss 3, do I get 2 cubed (=8) plus my 17 (20-3) bucks back?
The idea is that you don't get your stake back. The question is how much you would pay for a once-in-a-lifetime shot at winning an unlimited amount of money?
 
  • #15
PeroK said:
The idea is that you don't get your stake back. The question is how much you would pay for a once-in-a-lifetime shot at winning an unlimited amount of money?

Thanks. I was right:
gmax137 said:
I'm being dense
 
  • #16
Speaking of gambling … the Eurojackpot jackpot value is currently sitting at €120M. That’s such an extraordinarily large amount of money that it is silly. It is the GDP of a small country.
 
  • #17
Physicists discussing the non-spherical nature of a cow in a probabilistic vacuum. Love it.
 
  • Haha
Likes Tom.G
  • #18
Orodruin said:
It is the GDP of a small country.
If by "a small country" you mean specifically [Edit: the country with the smallest GDP which is] Tuvalu (GDP ~USD 60 million) then it is approximately twice the GDP of that small country. However the next smallest by GDP is Nauru (~USD 160 million) so it is not quite there.

€120M is a silly amount of money for a lottery win though. Also "silly" (or perhaps better described as "state-sponsored financial crime") is the income privately owned companies make from issuing tickets for the Eurojackpot, but we are getting far from the topic now.
 
Last edited:
  • #19
pbuk said:
If by "a small country" you mean specifically Tuvalu (GDP ~USD 60 million) then it is approximately twice the GDP of that small country. However the next smallest by GDP is Nauru (~USD 160 million) so it is not quite there.
I would classify Tuvalu anf Nauru as small countries so a GDP in between their GDPs would be the GDP of a small country. Obviously it is never going to be the exact GDP of an existing country.
 
  • #20
I'd suggest writing down the running total of what you would win or lose each time the coin is tossed, Because you do have to make a stake at each throw, in the original wording of the paradox (see wikipedia). I think the wording in your question is wrong.
 
  • #21
gmax137 said:
I guess I'm being dense today but how does "how much you're willing to pay" come into the game? Is it 1 buck for each coin toss? If I bet 20 bucks and heads comes on toss 3, do I get 2 cubed (=8) plus my 17 (20-3) bucks back?
I read it like this: You decide (up front) how much you want to pay. Then they start flipping the coin till it lands heads. Count how many times you flipped it, and they pay you 2^n dollars. I'm not sure if it lands heads on the first try do you get nothing or 2 dollars, though. It says "lasts n rounds". Does lasting mean that the rounds it lands tails?
 
  • #22
It took me awhile to understand, but now I think of it as the house asks the room "how much will you pay to play?" so it's kind of a bidding war to see who "gets the chance to play."
 
  • #23
scottdave said:
I read it like this: You decide (up front) how much you want to pay. Then they start flipping the coin till it lands heads. Count how many times you flipped it, and they pay you 2^n dollars.
Yes.
scottdave said:
I'm not sure if it lands heads on the first try do you get nothing or 2 dollars
Heads on first try is (in my opinion) one flip. ##n=1## ##2^1 = 2##. Two dollars.

Even if one decides that a win on the first round has ##n=0##, that would be ##2^0 = 1##. One dollar. That would halve the value of every finite outcome.

scottdave said:
though. It says "lasts n rounds". Does lasting mean that the rounds it lands tails?
I would interpret it as number of flips being the same as number of rounds. A first round win with one flip is ##n=1##. A fourth round win with three tails and a head is four flips so ##n=4##.
 
Last edited:
  • #24
jbriggs444 said:
Yes.

Heads on first try is (in my opinion) one flip. ##n=1## ##2^1 = 2##. Two dollars.

Even if one decides that a win on the first round has ##n=0##, that would be ##2^0 = 1##. One dollar. That would halve the value of every finite outcome.
Part of the paradox, of course, is that it doesn't matter. You might naively "expect" to win an unlimited amount in either case.
 
  • Like
Likes pbuk and jbriggs444
  • #25
  • Like
Likes Orodruin
  • #26
Gavran said:
“the mathematicians estimate money in proportion to its quantity, and men of good sense in proportion to the usage that they may make of it.”
Gabriel Cramer
Grigori Perelman turned down the $1 million prize for resolving the Poincare conjecture. So, maybe Gabriel Cramer was wrong about mathematicians. I think he's got them confused generally with chief financial officers and chief executives.
 
  • Like
Likes gmax137
  • #27
Gambling is a paradox but opposite to the St. Petersburg paradox.
 

FAQ: How much would you pay to play this fair coin flipping game?

How does the fair coin flipping game work?

The fair coin flipping game involves flipping a coin and guessing whether it will land on heads or tails. If the guess is correct, the player wins a predetermined amount of money. If the guess is incorrect, the player loses the money.

What is the probability of winning in the fair coin flipping game?

The probability of winning in the fair coin flipping game is 50%. This is because there are only two possible outcomes - heads or tails - and each outcome has an equal chance of occurring.

How much money can be won in the fair coin flipping game?

The amount of money that can be won in the fair coin flipping game depends on the rules set by the game. Some games may have a fixed amount while others may have a variable amount based on the player's bet.

Is it worth paying to play the fair coin flipping game?

This ultimately depends on the individual's personal preferences and risk tolerance. Some may see it as a fun and entertaining game with a 50% chance of winning, while others may not want to risk losing money.

Are there any strategies to increase the chances of winning in the fair coin flipping game?

Since the outcome of a coin flip is random, there is no guaranteed strategy to increase the chances of winning in the fair coin flipping game. However, some may believe in certain betting techniques or superstitions, but these do not affect the probability of winning.

Back
Top