My company makes crystal oscillators and I design them. I've only tried simulators a couple of times, and at best they show you that an oscillator will oscillate. If you are lucky, it will tell you the frequency.
What it won't tell you is the load capacitance of the crystal that is needed. You can forget startup-time or pulling range (for VCXOs). The model of a crystal as a series combination of an inductor, a capacitor and a resistor, with as capacitor in parallel with the lot, doesn't simulate spurious response or overtone oscillation.
Oscillator start up is often thermal noise that is amplified by the active element and filtered by the crystal. That is very, very difficult to simulate, and needs a very small time resolution and a very good voltage and current resolution, and the simulation needs to run for the time of many thousands of cycles. In real life, 32.768 kHz watch crystals take up to about 3 seconds to start.
Simulators are good for digital circuits. They can be used for analogue circuits with a bit of care, but they are not designed for working with crystal oscillators.