Humans often face a dilemma when making a decision: they can either make the decision quickly but inaccurately or accurately but slowly. How do humans decide when to stop accumulating information and make the decision? Many studies have found that participants tended to maximize the reward rate, i.e., the ratio between accuracy and time spent on the task. Hence, the participants seemed to stop to maximize the reward rate. In this study, I tried to identify the participants' strategies to maximize their reward rate by implementing the common strategies as computational models. I then reanalyzed the data collected by Fiedler et al. (2021) and compared the fitness between strategies. In their experiments, the participants were presented with two stocks with different probabilities of rising or falling and asked to select the better stock. The outcome of the stocks was sampled for the participants gradually, and the participants could decide when to stop sampling and make a decision. The participants were incentivized to make as many correct decisions as possible but were only given a fixed amount of time for the task. Thus, they were encouraged to maximize the reward rate. By comparing the strategies, I found that half of the participants approached the task with an evidence accumulator strategy. Each sample accumulates evidence toward one of the two boundaries, and the decision is made whenever a boundary is reached. The rest of the participants used strategies akin to likelihood estimation or counting the number of different outcomes between stocks.