Inspired by numerous examples of similar chess training programs that give you an actual game position and ask the player to pick the best move in the situation, I put forth the idea of the tactics trainer for Pokemon. Let's face it, not all of us have the ability to get a mentor, and some of us really don't learn that well by ourselves. Enter the tactics trainer, which presents "problems" taken from actual ladder matches (possibly use logs saved on Pokemon Showdown, because the server would crash if it tried to save this sort of data) and ask the student to select the "best" move in the given situation. It is assumed that the "best move" is the one selected during the actual match, which serves two purposes: It allows for multiple-move combination problems cross-checked against the log, and it permits the various problems to be assigned a level of difficulty based on the rating of the player who managed to put together that combination. While this is far from a perfect assumption, the idea is that higher-ranked players are able to see and seize more nuanced advantages that can be leveraged into a win than lower-ranked players. One possible fix would be to have the problem difficulty be dynamic based on the percentage of people who gave the correct answer. (Another possible idea would be to have Technical Machine run quality control by determining whether the action taken in a presented scenario really was best before making it into a problem.) Of course, Pokemon isn't chess in that there's numerous RNGs involved. In such a case, hax events such as a timely flinch are portrayed in the problem as they were during the match, and to paraphrase one of Phil's protips, a hax win is better than a sure loss. This ensures that the problem plays out as the battle did. To return to the multiple-move example, if a student gets the first action correct, then the replay proceeds with the opponent's action, then it's on the student to make the right second move. Possible tactical motifs covered by the problems could include forcing switches, setting up or removing hazards, boosting, sweeping, prediction, weather wars, trapping, revenge killing, scouting, and stopping a sweep. Between all these situations, the student should be able to develop a good body of working knowledge. If nothing else, when you have a run of bad luck on this thing you won't make an ass of yourself by raging about it in the main chat.