Using replay data on CL ##-## from 2022 April 18 - 27 (so the front month then was CL 06-22), and using simple 1 minute and 1 hour charts, with only a single indicator (attached full source code, see replayrate.cs) --an indicator whose only job is the measure the replay rate versus wall clock seconds -- I see an exponential decline in replay speed as the replayed minutes and hours of replay unfold. This should be quite easy for your developers to replicate, using that code.
Pictures speak more than words. I've attached several. The peaks in the orange replayrate indicator line at the bottom are read as: this is the replay ticks processed per wall-clock second.
Comment