This is a bit embarrassing, but I forgot that I calculated our last FI number in early 2023 and I never re-visted it. I went back and updated my post on our plan, using a 2022 baseline. In 2022 (end of 2022 is closest to early 2023), the number was $4,500,000($4,721,360 in 2024). As of June 2024, that's about $4,721,000. But after some more detailed analysis, that number changed yet again! More on that later.
How did we come up with this number in the first place?
It's completely made up
It's impossible to calculate an amount of money that will last you until you die. There are way too many variables and variation! Our number is a financial fiction, but a useful one. A compass does not tell you all the turns and hills you will encounter on your journey, but it does point you in the right direction.
The tools
There are many tools that help determine a financial independence number and I've used most of them. From the most basic/rudimentary to the detailed and advanced, I have used:
ERN SWR Toolkit and ProjectionLab are my tools of choice.
ERN SWR Toolkit
A good starting point, although a bit intimidating. Read through the blog post which also has some video tutorials.
Pros
Free
CAPE based withdrawal rate, useful if you want to retire at a non-market high
Variety of asset class historical return data (small cap, value, etc)
Cons
Difficult to understand at first
Difficult to compare multiple plans or scenarios
Does not help with taxes or tax reduction strategies
ProjectionLab
ProjectionLab combines a good user experience with power and flexibility. It has a few limitations, but it's still relatively new and the lone developer is dedicated to improving it, recently quitting his job to work on it full time.
Pros
Much better user experience than all other tools
Very high flexibility
Detailed outcome analysis
Detailed tax analysis
Cons
Only has S&P500 data for stocks
No CAPE-based withdrawal strategy, difficult to use if the market is not at a high
Our new number
Since we are now about a year out from our planned retirement, I wanted to do a more detailed analysis to cover more scenarios and update our plan. I passed our ProjectionLab plan to a financial advisor who does gut-check reviews for free. He came back with a few suggestions:
Medical expenses in old age were too low. I changed them to be roughly $15,000/year per person by the time we reached 65. I also added long term care expenses of $120,000/year 5 years before death.
No withdrawal strategy, resulting in $100,000 early withdrawal fees. This is outcome dependent (I chose the worst possible historical scenario to test), but I added a roth conversion strategy to hedge against this.
I chose a nasty year to retire, 1966. Our plan would fail if history repeated itself (poor market, high inflation), but I'm fine with that!
Monte Carlo vs Historical
Having sat through many videos, discussions, and articles on the pros and cons of Monte Carlo vs other simulation methods, I've changed my thinking. While the past doesn't repeat, it usually rhymes. There is a lot of evidence that random Monte Carlo simulations are not as accurate as historical simulations. That is why I use Historical (random year repeat) simulations in ProjectionLab and why the ERN SWR Toolkit is useful even though it does not have a Monte Carlo analysis.
While our plan would fail in about 10% of historical scenarios, we also have quite a bit of flexibility and buffer:
We planned for high medical expenses, but it is unlikely that we will need to spend that much.
We planned for 13k in travel expenses, this is a luxury and we can reduce if needed.
We planned for 30k in childcare for two children until age 5. If we are both not working full-time, this will very likely be less.
We did not plan for liquidity of private equity or side income, both of which are likely.
On the other hand, these expenses could also be higher (complicated medical issues, children with special needs). How much longer would we have to work to cover these? Having played around with the numbers, we could easily work the rest of our lives and still not be 100% certain to cover all scenarios (what if we both get cancer tomorrow?). At some point, we have to decide if we're willing to accept the risk that we might just have to gasp go back to work, or die to avoid those medical bills.
Our new number is...
$4,800,000. This isn't much higher than our estimate from 2022, our chance of success is just 5% lower (90%).
And surprisingly, we just touched our FI number yesterday! We're planning to retire in May 2025, so we're right on track. Doomsday predictions aside, we're feeling pretty good about our plan. Hitting our number a little early is unexpected, but I suspect a reversion to the mean will occur soon enough. As long as we're on the general track in 2025, we'll be happy.