Does anyone know why time clusters change in the past, as well as in the future, as data is being updated? I can understand why they would change in the future but the previous points should remain the same.
What Time Template is being used in the chart?
If it were based on a fixed number of bars then as new ones come in older ones are discarded thereby changing the historical clusters.
You've hit upon one of my favorite proprietary studies in AGET. I don't think I'm qualified to answer your question in depth (hopefully Marc will respond to this thread), but here's my take on the study:
Time clusters do change with time, and I have no idea why (it's a proprietary indicator, and we aren't given the formula). But IMHO that's the beauty of the study. It is dynamic and changes with the ebb and flow of the market, issue, etc. I have found the best results occur when you optimize the study continuosly (a bit of a pain in the neck, but highly worthwhile from a profit standpoint). When you see an exceptionally high bar, be on the alert for a CIT.
I have the EOD version, and I don't know what version you have (perhaps the esignal version optimizes automatically).
In any event, my advice is pay close attention to those optimized signals.
Best of luck
To be honest, it has been a very long time since Time Clusters were first developed.
I think Alex hit on the best explanation when he said it may be based on a fixed number data bars loaded. I think he is correct. The values change based as existing data drops off would be my guess too. I am only speculating, as well, but this is my guess. I really think it is related to this.