At some point you need data of a higher granularity in order to more accurately telegraph the faster fractals exact FTT's with greater accuracy. In the past many tools such as 2m YM, STR-SQU, tick charts, etc. were used for this purpose. You won't always be able to pinpoint a BBT's FTT with 100% accuracy on a 5m chart. Sometimes they give us more than the minimum required to construct them; and there's a procedure for handling those cases just as there is with everything else. Maybe this is the problem you're beginning to encounter as your annotations become more accurate. Tried to address some of your questions on the image.