That's because you have the wrong perspective, you're doing the trees not the woods. It's like you're trying to invent a car, so you sit down and spend all your time developing really cool gaskets, then wonder why you never have a working car. A really shitty car that actually drives around is better than a non-working one that has really well engineered parts.
Make the woods.
EDIT: start with a smaller scope, first, for a finished thing. It doesn't matter if the thing is small. You just need to build things you can finish (in the first video they recommend scoping for a full thing in no more than one month). Then, you see what needs to get done to make that thing, and you schedule for those things. Then, you make your crappy thing and put it up on itch.io, and forget about it.
That 'build a complete universe' thing is the very trap that the videos warn you about not doing. Everyone wants to do that stuff, and everyone fails.
For example, why not build a simple Masters of Orion clone. You can use your planet-drawing code in that. But, you have to schedule how much of Masters of Orion you can build in 4 weeks. In the first week focus on being able to produce fleets (of ONE type of ship) and move them around, 'capture' planets, but keep it as simple as possible: visiting a planet counts as capture, and if two fleets encounter each other, there's a very simple numbers-based battle. In the second week, flesh out a ship-design system, third week, flesh out a colony/production/tech system. fourth week, factions/diplomacy/end game, polish and upload. Things like adding in a graphical mode for the battles, that should be a whole new project in itself, so treat it as such.
The above would be "MooClone v0.1 - Minimum Viable Product'. Done get bogged down getting any one system "perfect" - randomly scattering stars around is good enough for v0.1, but you could spend a whole month just perfecting a galaxy-generator if you wanted. It would be a waste of time and lead to abandoning another unfinished thing however. Another goal should playable demo at each milestone. e.g. focus on a playable prototype from day 1, but one that can be finished, and the goal is that at each milestone you're left with a game that's still playable and ideally finishable. If part of one milestone is a situation where putting this system in will break the game until later on, then leave that thing out and make it it's own milestone in a future development round.
You can then scope out another one month project which takes you to 0.2, but since you already uploaded v0.1 you could take a break and work on something else instead.