I've been looking at a COVID cases graph and realized that the normal growth/drop plot is actually misleading:
In the middle of this plot we can see a peak growth rate of 56% and a peak drop rate of 23%. However if we increase a variable by 56%, we would need to reduce it by (1-(1/1.56)*100 = 35%, not 56%. Or on the other side that 23% drop would need to be compensated by a 29% increase, rather than a 23% increase.
is there a mathematical concept for a plot that takes this into account and presents growth rates in a way that equalizes the visual size of the growth/drop sides, either by stretching the negative growth axis or shrinking the positive growth axis? My goal is to plot this kind of weighted graph in Matlab but I have no idea if this is something that already exists or if I need to code it myself to see if it makes any sense.
Your statement about $56\%$ increase and $23\%$ vs. $35\%$ decrease is only applicable if you do one increase and one decrease and then return to where you were. We have no evidence here that we returned to where we were. Also, the increases and decreases compound over time, so we are interested in successive increases/decreases instead of the peaks.
You might be interested in a plot where the vertical scale is $\log(1+y)$ where $y$ is the value plotted in your question. You still have the problem that successive increases or decreases add up, but one week of $+56\%$ will be as much above the zero line as one week of $-35\%$ is below it.