Usually the "model" is some bogus Excel projection of historical volatility and returns into the future. That is nowhere near enough.
The quantitative mistake is not looking at returns as a random process (i.e. not ending up with a monte carlo sim with your assumptions, not modelling returns/volatility/etc. as random processes, if doing that then not using non-parametric models, if doing that then not using enough data, etc.).
The biggest practical mistake is understating the effect of volatility. Once you start or plan to take income, volatility becomes very important. Thinking that you can ride out volatility, the de jure thinking today, exposes you to intolerable risk (usually at the point where you are not able to bear that risk i.e. when you can't generate income anymore).
The aim of any retirement planning should be robustness. There is some kind of insane irony (the financial world seems to specialise in this) that people are attempting to invest in a way that requires maximum robustness but are often using strategies that produce totally non-robust returns (i.e. 100% equity ETFs).
If you are going to try this: stop and consider whether you really understand what you are doing.