Perhaps some useful insight for a good strategy in the original problem can be obtained from the solution of the following simpler problem in which you start again with rolling six dice. You are allowed no more than six rounds to roll the dice and after each round you have to put aside at least one of the dice, but you never put aside more dice than necessary in order to reach your goal. Your goal is to have put aside a 2 and 4 within the six rounds . Let $p(0,r)$ be the probability of reaching your goal from the state in which you have neither a 2 nor a 4 and still $r$ rounds are allowed . Also, let $p(1,r)$ be the probability of reaching your goal from the state in which you have put aside either a 2 or a 4 and still $r$ rounds are allowed. These probabilities can be numerically obtained by calculculating absorption probabilities in a Markov chain with 13 states including two absorbing states. I found that the probability $p(0,r)$ has the values 0.9567, 0.8725, 0.6943, 0.4180, and 0.1389 for $r$=6, 5, 4, 3, and 2. The probability $p(1,r)$ has the values 0.9350, 0.8384, 0.6651, 0.4212, and 0.1667 for $r$=5, 4, 3, 2, and 1.