I salute your perseverance, as I didn't actually get that far into it, but does he actually seek to show the vaccine is no better than the placebo by ignoring deaths from the virus?
So what he does is demonstrate a cute (and as far as I can tell, fairly well known) effect where if you've got two populations, one of which is increasing in size while the other is decreasing, and neither of which is dying at a greater rate than the other, and if you do your "how many people from each population are dying every week" calculation rather naively without accounting for some reporting delays then you get an effect where it looks like the people in the population whose size is decreasing are dying at a greater rate than the people in the population whose size is increasing (basically because you're fiddling with the denominators). With the delay that he hypothesizes, the peak of this effect looks like about a fourfold reduction in deaths, and it drops off to being negligible once you get near stable populations. I haven't checked the working, but it sounds plausible enough.
He then picks some ONS data for non-covid deaths and demonstrates that you get a similar pattern, although this demonstration is at a level of "look, these graphs have different x axis scales but they're kind-of similar looking - that seems consistent with there being a similar delay!"
He then doesn't do anything to try to demonstrate that this accounts for the apparent effectiveness of covid vaccines, but puts words like "vaccine", "placebo" and "statistical illusion" prominently in the title and sort of waggles his eyebrows suggestively and leaves it to people with an agenda to push to do their stuff.
