I thought that maybe I had taken account of every team's position (at the end of the match).
Take the first 3 rows of your source table where 6 teams play on the same day. At the end of that day I've got rankings and points as follows:
1. Inverness C 3.0620007
2. Motherwell 3.0000005
3. Kilmarnock 1.0010006
4. Dundee 1.000001
5. Hamilton 0.0000008
6. St Mirren -0.0299999

None of the other teams have played, so they don't come into the equation (perhaps those teams all count as 0, in which case St Mirren would be ranked 12th?).

So taking those 6 teams as being all being 'higher' teams, the next day, 4 more teams (from the now 'bottom 6') play, with Dundee United ending up with 3.0930009 points, higher than all so far, so I'd expect them to be ranked 1st, no? You say, though, that these teams have to be ranked 7th or below ('if the bottom six played each other on the same day and there were no other matches played then that's 3 fixtures and rankings would run from 6[7?] to 12'), so Dundee United should be ranked 7? I don't get it.

I may be getting this all wrong since I'm no fan of football (is this football!? maybe it's rugby?) and have no knowledge of it, even less of leagues and rankings - the only matches I've watched are world cup finals, and not all of them, oh, and I went to see Tottenham at home once to replace someone who couldn't go. This could be an advantage because I'm just like Excel from that point of view.

Confirm for me some assumptions:
1. The points in columns HT Algorithm Points and AT Algorithm Points are cumulative over time.
2. That the most recent points value for a team is the points found in either the HT Algorithm Points column or the AT Algorithm Points column, whichever has the latest date.

I'll take you through the process/logic/algorithm I've used when you've commented on the above.
Meanwhile, could you look at my results and tell me the first time it goes wrong, and crucially, why, and what it should be?