enter image description here

My attempt:

$(0, 1/3, 2/3)$ works as a strategy that strictly dominated $L$.

To prove that there are infinitely many of them:

Let $A_n=(0,1/2+\epsilon_n,1/2-\epsilon_n)$ where $\epsilon_n=1/n$ for $n=1,2,3,...$. We can see that for $n > 10$, $A_n$ strictly dominates L. So, since $n$ diverges, there are infinitely many $n$ for which $A_n$ is a strictly dominant strategy.

I chose $n> 10$ since $2<2.5-5\epsilon_n \Leftrightarrow 0,1> \epsilon_n$ and $1/n<0,1$ for $n>10$. The other inequalities don't require $\epsilon_n$ to be restricted.

Would this be a correct way to prove this?

  • $\begingroup$ Sir, you are posting a lot of questions in a short time period. Perhaps try to work them out a bit longer yourself. $\endgroup$
    – Giskard
    May 8 '20 at 16:41
  • $\begingroup$ Also, you are posting very specific questions. These types of questions are unlikely to be of help to future visitors, which is the primary goal of SE sites. $\endgroup$
    – Giskard
    May 8 '20 at 16:41
  • $\begingroup$ Lastly, questions of the "Is this correct" type are a poor fit for the SE format. $\endgroup$
    – Giskard
    May 8 '20 at 16:42
  • $\begingroup$ @Giskard sorry, coming from Math SE there's a special tag named "Solution-verification", and there such questions are normal. (At least I assume so since no one told me otherwise). Also, I work on these questions quite a bit, and if I get stuck or need feedback, considering there are no answers in my textbook for these, this is the only site I can refer to $\endgroup$
    – Slim Shady
    May 8 '20 at 16:50

That is a fine proof. Though you probably want to show your last statement "The other inequalities don't require $\epsilon_n$ to be restricted". It is fairly obvious, but won't hurt to be super clear.

Perhaps a simpler proof would be that any convex combination of the two strategies already found will also work. That is $\lambda(0,1/2,1/2)+(1-\lambda)(0,1/3,2/3)$ also strictly dominates $L$ for any $\lambda\in(0,1)$. So, in fact, there are not only countably many strategies that strictly dominate $L$ but uncountably many of them.

(Idea of the proof:) Notice that this holds because strict dominance is given by a set of inequalities (3 in this example) so if you have two strategies that satisfy those inequalities, any convex combination of these strategies will necessarily also satisfy the inequalities since expected utility is linear in probabilities. I'll leave the formal proof for you to write, but hopefully the logic is clear.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.