Quote:
Originally Posted by stinkypete
I'm betting all in on rolls 2-10. It's the highest EV strategy and I'm happy with the risk/reward on all of the bets. That's my final answer. I don't think any other answer is correct.
I would say there are some other answers that are equally correct and could have some advantages, like:
Bet all on same side every roll
It is simpler solution - actually, as simple as it can be: pick whichever side you want (head or tail), then bet all on that side in every roll from 1st to 10th. Beside being simpler, it also result in higher payout amount after win ( although with same EV, so wins would be twice less often )
And it is fairly easy to show that it reduces to same solution as yours. Assume EV=evA in your "bet 0 on first roll, then bet all in rolls 2-10 on whatever dropped on first roll (heads or tails)", then:
1) if we expand your solution to bet all on first roll (on any side, head or tails, select randomly), we will win in 50% (increasing our starting amount to $40), or lose in 50% (and stop our rolls). So, if we won first roll and continue to implement your solution, our EVwon = 2*evA (due to $40), but since there is only 50% chance we will win, our actual EV= 50%*2*evA == evA ... in other words, exactly same as for your original approach
2) since we will always continue to bet on same coin side (head or tails) in our 2nd-10th roll as in that 1st roll, it practically means if we randomly pick side and bet on it in all 10 rolls, we get same EV as in your approach
Actual EV, assuming p=probability for biased coin to drop on one side , is :
EV= $20*2^9*(p^10+(1-p)^10)
That part with (1-p)^10==q^10 is for rare cases when we select wrong side and manage to roll it 10 times anyway. If any coin bias is equally probable (unnecessary assumption, but just for giving example), than integral(p=0..1)(p^10+q^10)~0.18, and
EV~$1850. Anyway, EV is identical for your or mine solution, although it can be expressed differently:
a) in your solution, EV= $10240 with chance (p^10+q^10) ( or ~18% on average)
b) in my solution, EV = $20480 with chance (p^10+q^10)/2 ( or ~9% on average)
Since any solution with same EV is not clearly 'better' and will depend on subjective risk vs reward preferences, my point was not that my solution is better than yours - although higher payout may be important in some cases, and there are certainly solutions on other side of spectrum, with lower payout but higher chances than yours. My point was rather to object to your "I don't think any other answer is correct" statement. If you have used "better" instead of "correct", that could have been acceptable under 'subjective risk vs reward opinion' ... but "correct" has fairly precise meaning, and any solution resulting in same 'best' EV should be considered 'correct'.
Last edited by lostme; 06-04-2019 at 02:56 PM.