Suppose you play a game in which $n$ stacks of money each of different values, $a_1, ..., a_n$ are in a sequence on a table. You get to choose either the left-most or right-most stack, then your opponent can do the same, and so on, until all stacks are gone.
I need to prove that player 1 can always win at least half the money available when $n$ is even. I initially tried induction but the inductive step wasn't clear.
Suppose that you know the claim is true for $n=2k$ and consider a sequence of $2k+2$ stacks. Picking the maximum of the two end stacks will not always result in an optimal choice. If for instance the stacks are 1, 2, 100, 3 then as player 1 you might greedily choose 3 but then player 2 can choose 100. If you instead selected 1 then player 2 would be forced to select a sub-optimal stack and you get the grab the 100.
But I'm not sure how to prove this in general. Certainly if the stacks are $a_1, ..., a_{2k+2}$ and if $a_1$ is maximal then it's optimal to pick that, and likewise for $a_{2k+2}$. But there are also examples where $a_2$ is maximal and yet it's not optimal to pick $a_{2k+2}$. For instance 3, 4, 3, 1. The optimal selections go
Player 1: 3
Player 2: 4
Player 1: 3
Player 2: 1
Then player 1 has done better than trying to force player 2 to reveal the 4 stack.
Player 1: 1
Player 2: 3
Player 1: 4
Player 2: 3
One could theorize that if the max is $a_2$ or $a_{2k+1}$ and if that plus $\min\{a_1,a_{2k+2}\}$ is bigger than $\max \{a_1,a_{2k+2}\}+\min\{a_2,a_{2k+1}\}$ then you play the "defensive" strategy. But after playing it, there's no guarantee that player 2 will reveal the max on the next move. So I'm getting lost in the proof.
First player can ascertain to obtain $a_1+a_3+a_5+a_7+\cdots$ or $a_2+a_4+a_6+\cdots$, whatever they prefer.