Hi @MaxMarschall,
For Annual Daylight, and other rfluxmtx/rcontrib-based simulations, you generally want to adjust -lw with -ad. In the other topic you learnt about the HB Radiance Parameter component. If you look at the value for -lw you will see that it is set at 1/(ad*100) for the different detail levels, which should likely be low enough to avoid splotches in most cases.
I believe that the reason why you don’t see a difference in the results with higher ambient bounces is because the rays at that point no longer contribute significantly, or they are terminated at that point. Every case is different but I imagine that you will not see much difference past 5 ambient bounces for your simple model.