Percentage of change, compared to a certain number of candles back, and optionally smoothened as a moving average. I use this for Covid-19 numbers, where I compare to 7 days ago, and use a 7 day moving average over that, which gives me a nice idea of the real corona trend.
In the example, I look at the active Corona cases in Switzerland, ie confirmed cases minus recovered minus deceased. I compare to the numbers from exactly 7 days ago, as numbers often spike around the weekend, and then put a 7 day moving average on that. I see lower highs, which is good, although negative change would be better of course...