sneaklyfox
Legend
Marcel, I think many people would do better doing less timed solves (like me), but I think you would do better doing more timed solves. While slow solves help with things like look ahead, F2L efficiency and reduced rotations, I only find that push to increase speed when doing timed solves. I don't know if this is the case, but it's possible that you've reached a state of complacency. Because you do so few timed solves, most of your practice is spent going at a steady pace that is comfortable to you. I think you may need that timer going in the background to force you out of that comfort zone and get you to solve faster. I'm not sure if this will actually help, but given how you practice this is what I would add. Your large number slow solves without a lot of timed solves might actually be training you to solve at a slow pace.
Sounds plausible and I would agree with jayefbe that you should do some more timed solves. I hardly ever do untimed solves unless I'm not at the computer (but I usually am). Sometimes I do fast solves and sometimes I do "slow" solves (still timed). When I do slow I focus on smooth flow. I note the time even though it normally takes a few seconds slower. Sometimes I'm surprised that my slow solves still get good times but I guess that's because instead of moving quickly I eliminate pauses instead. PLL time attacks are fine but I'm actually more in favour of just training one PLL at a time. Aim for faster turning for this so your fingers get used to the speed. F2L should gradually become faster and then you just need to worry about look-ahead catching up to the turn speed.