You most certainly can. First you'd have to address the limiting beliefs and viewpoints that inform your perception that you cannot. Perhaps not in the manner that's it's accomplished on this thread. To start, one would have to begin by logging bars, then various aggregates of those bars and the relation to volume as it increases and decreases on a per bar level as well as the aggregate level. Moving averages would be off the path, one would need to retain the basic granularity of OHLCV. It would be a different representation and visualization of data from the more common PV chart. Creating a stop log would support the effort. Price retraces stops until it doesn't at various levels of observation as it continues the dominant trend. Then when the trend fails, the market moves with symmetry in the opposite direction until it fails and the cycle repeats. The repeating cycle logic can be coded with VBA and excel but it would be helpful if one also incorporated a sql db to store the data. It's easier to use EOD data to start and work in a batch mode. Over time one then builds a db of reference that illustrates all the above to be true. You just have to make it true for you regardless of someone's else's failure or lack of due diligence. Your only limitations are the ones you self-impose.
Sounds like you have done something similar before? It is probably too complicated for me to program.
just found this thread, super interesting. One thing that strikes me is why would images of candles produce better results than say the actual time series of bars? Seems like the image processing stuff would introduce unnecessary noise, no?