The use of online judges has become popular as a tool to improve programming skills by measuring the efficiency of solutions and checking their compliance with problem requirements. However, when a solution exceeds the time limit, the judge provides little information about the cause of the error. In order to improve the information in the verdicts provided by judges, this study proposes to extend its functionality by implementing a clue module capable of informing users about the cause of the obtained verdicts, focusing exclusively on time-limit exceeded or TLE verdicts. To achieve this goal, we propose to classify TLEs into three categories: infinite loops, wrong approaches and suboptimal solutions. The module classifies the solutions by making use of a part capable of predicting the order of complexity of a solution sent to the judge, timing its execution time for a series of test cases. For this part there are two implementations, one based on artificial intelligence models and the other based on regression functions. This work aims to improve the user experience when interacting with online judges, without replacing them, by increasing the amount of detailed information provided by the verdict system on the online judges.