Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Optimal number of goes

  1. Aug 21, 2007 #1
    A game has two outcomes win or lose, the probability of winning is 1/3 and
    the probability of losing is 2/3, if i win i will receive 3 units but if i
    lose i will lose 1 unit, I've calculated that the average rate of return on
    this game is 33.3%. Over time it becomes more likely that i will receive my 33%
    return. My question is if i play the game 1000 times which go would give me the best bang for my buck in other words, what is the optimal number
    of times i should play to get the best value for money.

    Ps. Any help would gratefully be appreciated,my thoughts are that a graph with probability of gething 33% return on the y-axis(scale 0 to 1) and No of goes on the x-axis would produce a sigmoidal like fuction with the optimal number being at the point of inflection but how do i find that point!
  2. jcsd
  3. Aug 21, 2007 #2


    User Avatar
    Science Advisor
    Homework Helper

    In general, playing more times will give you more money. After 30 plays your expected earnings are 10; after 3000 they are 1000. The problem becomes more interesting if you have a certain amount of starting money and can't go negative. For example, if you start with 1, you have a 2/3 chance of going broke after your first try.

    What do you mean by this?
  4. Aug 22, 2007 #3
    The Sigmoid like function i referred to can be seen
    here(http://en.wikipedia.org/wiki/Sigmoid_function )

    If 1 on the y-axis represents certain probability of us
    getting our 33.3% return.
    Then as the function tends toward 1
    it becomes more likely we shall receive our 33.3%

    I apologize the sentence should have read
    Over Number of goes it becomes more likely that i
    will receive my 33% return.
  5. Aug 22, 2007 #4

    matt grime

    User Avatar
    Science Advisor
    Homework Helper

    No, that still doesn't really make sense. I think you're attempting to invoke some rule that says in the long run things tend to the expected value. Assuming you can lose arbitrarily many times, that is true. Some of us might go broke in the interim.

    Your question is still ill-posed. You assert you're going to play 1,000 times and then ask how many times to play. That makes no sense.

    At each stage you expect to gain, so play as many times as the sucker who's running the game will let you.
    Last edited: Aug 22, 2007
  6. Aug 22, 2007 #5


    User Avatar
    Science Advisor
    Homework Helper

    Why do you care about what average return you get? You just want to maximize total winnings, don't you? That is, you'd prefer to make 100 units in 300 plays (average return: about 33%) than 10 units in 10 plays (average return: 100%), right?
  7. Aug 24, 2007 #6
    The question came in three parts and was verbally stated so
    please excuse my poor interpretation, i will try to provide more clarity

    Part1 was to work out average rate of return, given the stated stakes and probabilities, which i worked out as 33.3%.

    The second part was to draw a graph with probability of getting the 33.3% return on the y-axis and number of times played on the x-axis
    This graph tend towards one the more you play and looks similar to
    the wikipedia sigmoid function that is linked in my 2nd post.

    Part3 was to find the point on the graph that gives the
    player the best value for money i was informed that this
    is the point of inflection.

    I cannot do part 3 and am unsure if i need the information from
    part 1 to find the point that is why i stated the question as such in my first post thank you very much for replying, I truly appreciate it.
  8. Aug 24, 2007 #7


    User Avatar
    Science Advisor
    Homework Helper

    I still don't know what you mean. Is this the chance of getting at least 1/3 return, or of exactly 1/3?

    Tries - At least - Exact
    0 - 1 - 1
    1 - 1/3 - 0
    2 - 5/9 - 0
    3 - 7/27 - 0
    4 - 11/27 - 0
    5 - 17/81 - 0
    6 - 233/729 - 0
    7 - 379/2187 - 0
    8 - 1697/6561 - 0
    9 - 6883/19683 - 448/2187
    10 - 4195/19683 - 0
    11 - 17065/59049 - 0
    12 - 31483/177147 - 0
    . . .
  9. Aug 24, 2007 #8


    User Avatar
    Science Advisor
    Homework Helper

    How do you know it has an inflection point? A convergence result would look like an "r" shape, rather than an "s" shape, IMO.
    Last edited: Aug 24, 2007
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?

Similar Discussions: Optimal number of goes
  1. A number (Replies: 4)

  2. Optimization Question (Replies: 0)