For a long time I am debating with myself how to approach the “Close” problem. I can summarize it very simply: When using any of the Open-High-Low-Close prices, in our testing, and algorithms, we are using a thing that do not exist in real life. In real life, these numbers are known of course, only in the next bar or only in the last tick of the current bar. So, if the logic depends on doing something, intra-bar, by the status of these numbers, it is only in the last tick, (and therefore with the possibilities of suffering a very extended bar) or to enter and exit intra-bar, in case these numbers changes as to change the decision to enter or exit. The two alternatives will create a very harmful effect on the system performance. I am aware of the possibility to create a buffer around the line of decision, and so, not to enter-exit back and forth, too much, but then the question will be how wide this buffer should be, relative to the time-frame (and volatility). Any ideas or suggestion? From listening avidly to your audios, I think that you deal with a more “long-term” approach (weekly bars) but I think that this “problem” is relevant in all time frames and even in all price representations. Many thanks in advance.
Nothing I prescribe is intra or day trading. Simplest way I can offer feedback. This is not a trend following concern.