The trading rule is simple: generate signals from news, enter at the next day open, and exit within the next two trading days. However, the test window in the paper is short, mid-November 2024 to late January / February 2025. So the reported Sharpe is not evidence that the method is robust.