That's what I've always assumed it does since back when quicktime player barely even ran on my PC yet for timeline operations it was significantly more responsive than WMP/MPC.

For Losslesscut I just get around this by encoding my input from source using keyint=n:scenecut=0 in ffmpeg where n is a manually set keyframe interval.

So e.g. if my expected cut occurs on a frame that occurs at t+10 seconds of footage, n can be the same as the fps and then there'll always be a keyframe exactly at timestamp 00:00:01, 00:00:02 and so on. I can then open it in losslesscut and easily snap to the frame I want and make the cut losslessly.

Yeah the first encode generally means a lossy transcode by the time I get to my final video but being realistic that'd be a part of my workflow either way and this way it's less