Overview
The greatest challenge for retirement planning is forecasting asset returns. Workers save their money by buying assets, which may include stocks, bonds and/or houses.
There are no easy ways to forecast asset returns. A standard method is to assume some model of the assets time evoluation (for example, geometric brownian motion for equities) and then to take the average historical return and volatility as the model parameters.
These types of models tend to be unrealistic when played out over long periods of time. That is, the GMB model can include unreasonably high asset prices in its list of possible outcomes. This may not be an issue, especially when the analyst is just calculating average outcomes. However, there are methods that will force some degree of mean reversion in the asset growth, such as the Brownian Bridge model.
Simulations
The mathetical complexities of deriving a portfolio distribution given a set of assumptions about random asset returns can be daunting. The typical way to deal with these challenges is to encode the assumptions into a simulation and then running the simulation many times in order to build a distribution.