1. Work solo, unless you can pay people.

2. Pick a project you can safely make solo in a reasonable time.

3. Make detailed documentation and schedule before you start.

Like most have already said, finishing a game - or anything for that matter - requires a lot of patience, determination, and strong will. You need to be dedicated and willing to push yourself to work on it, sometimes even when you don't feel like it anymore. Starting a project you always have a lot of drive and determination, which tends to fizzle out as you go along. You need to fight against this.

You also have to accept a point when you and your team say it's done. That is another important thing. Perfectionism is often a killer of completion. You want it to be functional and as perfect to what you have in your mind as possible, this is true, but once you get to that point and are still fiddling with minor details......it might be time to consider it finished and move to the next step. Otherwise you can fiddle for years.

"Software is never finished, it is only abandoned."

Where is this a quote from?

"Software is never finished, it is only abandoned."

Where is this a quote from?

Art is never finished, only abandoned is a quote from Da Vinci. Coding is an art form (think of range-for, so beautiful)?

The best advice I can give is to never "fix it later". Especially with hobby projects, it's easy to hold off fixing annoying bugs in favor of the "fun stuff", and that invariably makes fixing it all the worse. Spending lots of time not adding features is incredibly demoralizing for a hobby project, so try to amortize it out as much as possible.

You're still going to have to do a lot of random bugfixing near the end, but try to make sure your game is always in a state where everything you've done previously still at least *seems* to work.

In a professional context, we effectively get around this by having other people tell us what to do (bug reports, essentially) and being paid to fix them.