My Excel skills are not that good.
About 2-3 years ago someone did it for me and my wife.
He included RMD, MAGI, deductions, Fed and state taxes.
The assumption was 6% annual performance.
He already proved it at that time, but I didn’t listen
We had 6 scenarios, but only 3 really matter.
1) High conversion of $300K. Effective tax about 24-28%
2) Medium conversion of $165K. Effective tax 22-25%
3) Lower conversion of $70K. Effective tax 12-15%
At age 85 final numbers in millions
1) High conversion. Taxable=0….ROTH=6.8…TIRA=0. ..............Total portfolio: 6.8.....Total tax=$729
2) Medium conversion. Taxable=0.8….ROTH=4.6…TIRA=1.5...Total portfolio: 7.0.....Total tax=$769
3) Lower conversion. Taxable=2.0….ROTH=2.1…TIRA=3.0........Total portfolio: 7.1.....Total tax=$1658
At age 90 final numbers in millions
1) High conversion. Taxable=0….ROTH=10…TIRA=0...................Total portfolio: 10.....Total tax=$729
2) Medium conversion. Taxable=1.3….ROTH=6.3…TIRA=1.5....Total portfolio: 9.1....Total tax=$1156
3) Lower conversion. Taxable=3.2….ROTH=3.1…TIRA=2.8.........Total portfolio: 8.1....Total tax=$2109
The high conversion looks the best.
The key factor is my investment style — all of my trades are short-term. I may hold positions for 2, 6, or even 18 months, but since 2018 I’ve never held anything for more than a year. When market risk was high, I sold and stayed out until conditions improved, then bought back in.
This is why increasing my joint/taxable account is a major concern for me.
The figures above don’t reflect short-term trading activity, which results in a higher tax burden.
The medium and lower conversion scenarios would therefore produce worse outcomes.
At 8% growth the numbers would be much higher. My own numbers show about 10 million for the high conversion at age 85 instead of 6.8 and better than the lower ones because of taxes.
I'm done with this exercise.
In scenario 1, you end up with some high tax rates while you do conversions and then low tax rates after you are fully converted to Roth. That is pretty much guaranteed to be non-optimum as in those later years, you end up with unused 12%, 10% and perhaps 0% brackets. Meanwhile you paid 24% + IRMAA surcharges, NIIT, lost OBBBA deductions, etc.
With $2.1M in tax deferred and aggressive growth assumptions, converting to the top of the 24% bracket or nearby IRMAA tiers could be defended. In your case, that may especially true since your in-and-out approach generates high capital gains and paying for Roth Conversions out of taxable will help extinguish that tax drag by running the taxable account down. But I doubt the math works out if you do so many conversions that you exhaust the tax deferred account. At some point in the conversion process you've brought the tax deferred balance down enough that continuing to convert means converting at high tax rates and avoiding future lower tax rates, which can't be right.
Note that neither metric that you looked at to evaluate conversions is really what you want. Looking at total taxes paid in a time series where there is growth above inflation is always a misleading metric and will always favor aggressive early conversions, but it says nothing about how much you end up with at the end. Neither can you just sum up the account values, you need to tax correct them. The big adjustment is that any tax deferred that is left at the end of plan has an embedded tax liability that you should account for.
Folks also often leave some unconverted amounts in tax deferred as a way to pay less dearly for long term care. Let's say you are 85, with 100% in Roth and now need long term care. That care would be tax deductible, but the deduction is wasted as you have insufficient income to use it.
Another factor to consider is that if you really end up with $15+M, you may decide to give some to charities and doing so from tax deferred via QCDs and bequests is the best kind of money to give as it's never been taxed.
There are inexpensive tools (~$120/year) in the marketplace that can help you with this analysis. It seems like a waste of time to try to build your own tax model of sufficient complexity to model all the interacting pieces here. Even if you built something on your own, you've no good way to validate it, it will have errors and omissions.
.
