The acquisition algorithm uses a two-stage selection process that first picks a Rank Factor Class via a freshness-based diversity weight and then applies Thompson Sampling, Upper Confidence Bound (UCB), or epsilon-greedy policies to choose the most promising hypothesis.