This is a great question, and is something that annoys me a bit, as well. (read to the end to see a list of links)
Stellar evolution is *super* complicated. We understand the basics of it quite well, and for a lot of stars, our models do a really good job of matching up to measurements of real stars. However, it's very hard to do this right. First, we don't have perfect data. We can't directly measure the mass or radius of a real star. And even something like the temperature of a star isn't always trivial to measure. So lots of the data are estimated. Further, we can't actually observe the evolution of a single star (well, we can watch it, but so far, we've only been watching for maybe a few decades or so, depending on when you consider our technology to have been good enough to do any of this). So we have to presume that some stars are similar to the way that other still will be in the future. It's like looking at a whole bunch of people, and guessing how you will age by seeing what old people look like right now. It's possible, but not exact, and how you age depends a bit on what you eat, what happens to you, etc. Similarly, we can guess what the sun will do, but it has slightly different properties of the stars we think it will look like.
One really big issue is what astronomers call 'metallicity'. Basically, what is the relative fraction of elements in a star? How much iron is there, relative to hydrogen, etc?
Finally, we can't see inside a star, which is where most of the action happens. So this is all based on physics calculations. We're good at that, but not perfect.
Ok, now on to how this directly relates to Universe Sandbox. Since stellar evolution is so hard, we let the full-time professional astronomers compute the models. These are called 'isochrones', in general, and we've adopted a suite of them, so we have models telling the us the temperature, radius, luminosity for a range of stellar masses and ages. And we turn that data into what you see.
Unfortunately, the isochrones, although quite good and pretty accurate, don't give us predictions for very low-mass stars ( less than 10% of the Sun's mass), or very high-mass stars. And they don't give us predictions for these longer times. (recall the age of the Universe is a bit less than 14 billion years).
So, this is what we're stuck with for the time being. In the future, I'm planning on incorporating some other isochrones that are focused on lower-mass stars (where the physics is a bit different, since the stellar atmospheres are very different - at low temperatures, molecules and even clouds can form in the stellar atmosphere - so you need different models), higher mass stars, different metallicities, and for a longer timeframe.
This is a work in progress, but we'll be making it better and better as we go.
http://en.wikipedia.org/wiki/Stellar_evolutionhttp://en.wikipedia.org/wiki/Stellar_isochronehttp://en.wikipedia.org/wiki/Metallicityhttp://stev.oapd.inaf.it/cgi-bin/cmd