This isn’t the first time I’ve seen repetitive patterns, not from deep history, but from the brief past. Here is one example:
Watch how similar the green patterns are. The time difference between them is only a week and a half. Both green up-trend patterns ended up as down-trend patterns. We could have detected the similarity of the green patterns by eye and opened a short position at the end of the second green pattern. The question is: how do we do that automatically? I have already tried a probability network. The problem with that network is that it is very sensitive to pattern distortions. For example, two red patterns seem to be the same to the eye, but the second one is almost twice stretched out in time with respect to the first one. The movement along the price axis is also of different amplitude. If we try to take the vertices of the zigzag, one pattern will have more vertices than the other and the net will get jammed. Thoughts?