1.) Try to make the artwork in the game look as good as possible. Just because the calculator isn't very powerful doesn't mean you can't make a good-looking game.
2.) Don't release a public version of a game that is known to have significant problems (beta releases are different, though).
3.) Ask for advice from others on ways that you might be able to improve a game. Often times you may be biased in the judging of your own programs and they may not be as great as you yourself perceive them to be.
4.) Try to make something that will be not just played and deleted within a brief period of time. Aim to make the game something great.
5.) Optimize, optimize, optimize! Make it your goal to make the game code as efficient as possible.

Also, even though this does not make the game itself any more professional, it would be a very good idea to distribute the program under the GPL and include the source code with the game (like DJ said- the TI community is very open source, and I know I for one agree with this unofficial philosophy).

I have to disagree with you on #4, Scout. I can't speak for others, but I think application games are generally a waste of app pages. You don't need to install a whole app for a program. Things like Wolfenstein are very impressive without them.