Your table is very hard to follow, you're putting most of the visual encoding into stuff which isn't carrying data. Tufte's concept of data ink is useful here.
Because you've got so much non-data-ink it's very hard to scan your table and get a sense quickly of what the current state of play is.
If you're going to use color to encode information (green for good, red for bad), you need not to be also using it as a design element to visually separate bits of the table. You absolutely need not to be using the same color (green) in a data ink role as a "good" indicator on numbers, and also as a design-ink role to visually separate the labels row.
I'd suggest paring back the colours on the chart itself, and instead just use them to highlight good/bad data points. You can easily distinguish the labels row with a bottom-border, instead of using a colour and then confusing everything by also using that same colour as part of the data encoding.
The other thing that would massively improve your table would be to leave cells that have no data (i.e. cells for future dates) blank. It's currently hard to read because you've got all those red 0s for days that haven't happened yet. You could just leave those cells blank until filled, and everything would be much more readable.
Here's a quick picture showing how I'd improve your table:
There's probably something else you can do with the way you're splitting out your data:
Having two numbers in the same cell is very rarely the right approach.
Unfortunately since I don't actually understand what these numbers represent, I can't guide you a better way to present them.