Show HN: Retirement Calculator/Simulator | Hacker News Reader

Show HN: Retirement Calculator/Simulator

lunchmodel.com

11 pointsdnadler3 comments

Hi everyone -- I've been working on this for a while in my free time. It's not 'feature complete' yet, but I'm eager to get some feedback.

It's basically a monte carlo simulation based on market characteristics (correlation, volatility, etc.). The 'killer feature' is that it can accurately account for granular cash flows throughout a person's life. For example, selling or buying a house, receiving an inheritance, a recurring salary, social security, etc.

This page has some more info about how to use it, and how it works: https://lunchmodel.com/learn-more. I'm happy to answer whatever questions you have, too.

Try it out, and let me know what you think. It's running on a small server, so hopefully it doesn't get too popular :).

One important note is that the form's state is saved in your browser's localstorage, which means you can leave and come back. If you want to delete this data, you can click the 'reset' button at the bottom.

Some features I'd like to add are:

- How would I have performed in XXXX period?

- Save and Load state

- Some nominal registration fee ($2? What would you pay? Anything?) to unlock the 'advanced features'

- What did Equity/Bond performance look like in the 99th and 1st percentile simulations?

The problem with granular simulators is that the results depend on unknowable parameters. Do you compute a robust model the accounts for the 90% confidence interval of those parameters?

The problem with Monte Carlo is the it tells me I'll die with somewhere between -$5M and $20M.

It generates returns for the market based on the covariance matrix and returns assumptions. These do fluctuate, but long-term estimates of these values are pretty well defined. The simulation here generates several thousand random returns based on these assumptions, and the results indicate the distribution of possible outcomes.

The results you are seeing suggest that under these assumptions, there is a 1% chance that you will have -$5M and a 1% change that you will have $20M at the end of the simulation. (1st and 99th percentile outcomes).

With different parameters (less savings, lower income, etc) you may have a much high likelihood of running out of money.

Knowing what this distribution looks like can be quite informative.

I could modify the market returns generator to allow for variance around the historical market assumptions, or provide some sensitivity analysis. On the whole, though, that will simply expand the distribution of the final results. I believe it is better to target more conservative assumptions than it is to include both conservative and aggressive assumptions. If you know that your conservative scenario still plays out well, then you can be confident that your plan is sound.

I guess, to sum up: the fact that monte carlo returns a distribution is a feature, not a bug. You can't know the future for sure, but you can put some bounds around the likely outcomes. If you have a 10% of running out of money (for example), you should probably save more.

Reply on news.ycombinator.com