What a Tax Optimizer Actually Does (and What It Can't)

The four levers a tax optimizer pulls, the inputs it needs, and where it stops being useful.

Last updated: June 24, 2026

"Tax optimizer" appears in marketing copy across financial planning tools, but the term itself isn't a regulated definition. This article describes what a tax optimizer is doing under the hood โ€” and equally important, what it isn't doing.

NIIT trigger (MFJ, 2026)
$250,000
+3.8% on investment income above
0% LTCG cap (MFJ, 2026)
$96,700
Tax-free realization room below
IRMAA tier-1 cliff (MFJ, 2026)
$218,000
Medicare premium jumps above
NIIT thresholds last indexed
2013
Bracket creep is real

What "tax optimization" means

Tax filing is annual and backward-looking. It computes what you owe on income you already realized. Tax optimization is multi-decade and forward-looking. It models how the order, timing, and account-source of future income affects your lifetime tax bill, and it suggests adjustments while you still have time to make them.

A typical retirement-tax-optimizer answers questions like: Should I take this year's spending from my Traditional IRA, my brokerage, or my Roth? Should I convert $40K to Roth this year, or wait until I'm 65? If I take Social Security at 67 instead of 70, how much more tax do I pay over the next 25 years?

The four levers a tax optimizer pulls

Four levers a tax optimizer pulls each yearYear-by-year, an optimizer pulls four levers1Withdrawalsequencing
Which account does this dollar come from?
2Rothconversion timing
How much to convert this year vs later?
3Capital-gainsrealization
Sell into 0% LTCG bracket while open?
4IRMAAmanagement
How close to the cliff this year?
Decisions in one lever change the optimum for the next.
Figure 1 โ€” A tax optimizer solves the four levers jointly, not in isolation. A Roth conversion decision (lever 2) shifts the next year's IRMAA exposure (lever 4); a capital-gains realization (lever 3) can crowd out conversion room. The interdependence is what makes a tax optimizer worth running.

1. Withdrawal sequencing

Which account does each year's spending come from? The traditional "taxable โ†’ tax-deferred โ†’ Roth" rule is one heuristic, but a tax optimizer evaluates per-year alternatives. Drawing more from a Traditional IRA in a low-income year fills lower brackets cheaply; drawing from Roth in a high- IRMAA-risk year keeps MAGI under a cliff. Optimal sequencing is rarely the same as the textbook order.

2. Roth conversion timing

A multi-year Roth conversion strategy fills lower brackets each year before RMDs arrive. The optimizer decides how much to convert in each year, balancing today's marginal rate against the projected rate when RMDs become forced withdrawals. Done well, it can shift significant pre-tax balances to Roth at meaningfully lower lifetime cost than waiting.

3. Capital-gains realization timing

Long-term capital gains have their own bracket structure (0%, 15%, 20%). The 2026 0%-rate cap is $48,350 taxable income single / $96,700 MFJ. A tax optimizer identifies years when your total income leaves room under the 0% cap and recommends realizing long-term gains to reset cost basis tax-free. It also coordinates loss harvesting and avoids stacking large realizations against Roth conversion years.

4. IRMAA threshold management

Medicare Part B + D surcharges (IRMAA) jump at cliffs โ€” not phase-outs. Crossing the 2026 tier-1 threshold of $109,000 MAGI single / $218,000 MFJ by even a single dollar triggers an annual surcharge per person. A tax optimizer models the MAGI consequence of every withdrawal and conversion decision and warns when you're crossing a cliff for marginal benefit. For a worked example of how IRMAA stacking impacts an HNW Roth conversion strategy, see the Golden Window article.

Marginal rate stack as MFJ taxable income rises (federal + NIIT)Marginal rate stack as MFJ taxable income rises (2026)Federal bracket + NIIT compose on the next dollar of investment income0%10%20%30%40%Marginal rate on next $1$100Kunder all triggers22.0%$240Kstill under $250K NIIT24.0%$300KNIIT triggered above $250K27.8%$420Kjumps to 32% federal bracket35.8%Federal bracketNIIT (3.8% surtax)
Figure 2 โ€” On the next dollar of investment income, federal bracket and NIIT compose. NIIT kicks in at $250,000 MAGI MFJ and adds a flat 3.8% to every additional dollar of investment income above the threshold. IRMAA works differently โ€” it's a flat per-couple dollar surcharge that triggers at each cliff (tier-1 at $218,000 MAGI MFJ adds ~$2,328/yr per couple in combined Part B + Part D premiums), so it's not a marginal-rate percentage. A tax optimizer evaluates both โ€” the percentage stack on each marginal dollar AND the dollar cost of any cliff crossings.

(Honorable mention) 5. NIIT awareness

The 3.8% Net Investment Income Tax kicks in above $200,000 MAGI single / $250,000 MFJ โ€” thresholds frozen since 2013 and never indexed for inflation, so more households drift into NIIT each year. A reasonable tax optimizer models NIIT implicitly when it computes the effective marginal cost of additional investment income.

What inputs a tax optimizer needs

  • Account balances by tax type (Traditional / Roth / brokerage / cash) โ€” not just total net worth
  • Annual spending target (and whether it's inflation-adjusted)
  • Social Security claim age (or range of ages being evaluated)
  • Pension or annuity income, if any, with start year
  • Marital filing status and state of residence (state-tax bracket varies wildly)
  • Expected investment return and volatility (or a default the tool justifies)
  • Life expectancy (single or joint) โ€” affects RMD horizon and Roth payoff window
  • Charitable intent โ€” QCDs after 70ยฝ change the optimization meaningfully

Tools that ask for fewer inputs than this are taking shortcuts. That can be fine for a first-pass estimate, but it means the output is a starting point, not a recommendation.

What it produces

  • A year-by-year withdrawal plan that names which account each dollar comes from
  • A recommended Roth conversion schedule (per-year dollar amounts)
  • Cumulative federal + state tax paid over the modeled horizon vs. a baseline strategy
  • The lifetime tax savings of the recommended plan vs. doing nothing
  • Sensitivity: how the answer changes if expected return, lifespan, or spending changes

What a tax optimizer CAN'T do

1. Predict future tax law

No model knows 2032 marginal rates. Most tax optimizers assume current law continues; some let you toggle "TCJA sunset" or "rates rise to X%" scenarios. Take any tax-savings figure as conditional on the assumed legal regime, not a guarantee.

2. Override behavior

The optimizer can compute the optimal plan. Whether you follow it for 25 years is a separate question. A 5-year-cumulative-savings number assumes 5 years of disciplined execution that real households often don't deliver.

3. Know when you actually die

Roth conversions are most valuable when paid back with many years of tax-free compounding. They're least valuable if you die before recovering the conversion tax. The optimizer uses an assumed life expectancy; reality varies by 10+ years either direction.

4. Hedge against unmodeled cost shocks

Long-term care costs, sudden healthcare events, family financial obligations โ€” most tax optimizers don't model these. A plan that's optimal under modeled cash flows can become wrong when an unmodeled $200K cost arrives.

How to tell if a tax optimizer is reasonable

A quick credibility checklist when evaluating any tool:

  • โœ… Uses current-year IRS values (2026 brackets, IRMAA tiers, contribution limits)
  • โœ… Models IRMAA as cliffs, not phase-outs (the actual statutory behavior)
  • โœ… Runs a multi-year horizon, not just a single year
  • โœ… Surfaces assumptions (returns, volatility, life expectancy) and lets you change them
  • โœ… Produces results you can roughly verify by hand for simple cases
  • โœ… Tells you when its recommendation isn't actionable (e.g., "you're already optimal")
  • โš ๏ธ Cautious of: black-box outputs, no assumption disclosure, single-year-only horizon, no IRMAA modeling

The honest framing

A good tax optimizer turns a question that's impossibly large to evaluate by hand โ€” "what withdrawal sequence and conversion schedule minimizes my lifetime tax" โ€” into something tractable. It doesn't replace judgment, doesn't replace an advisor, and doesn't deliver guaranteed savings. It does deliver a defensible quantitative starting point for the conversation.

Praxion's tax optimizer models all four levers above with 2026 IRS values and per-year IRMAA cliff awareness. See the methodology doc for the inputs, assumptions, and limitations โ€” or run a free analysis against your own numbers. Praxion is a decision-support tool, not a registered investment adviser.

Related reading

The HNW Roth Golden Window โ†’
When the four levers stack against a $5M+ portfolio โ€” IRMAA, NIIT, surviving-spouse compression, and the 32% bracket strategy.
Portfolio Rebalancing: A Tax-Aware Guide โ†’
NIIT details ($200K / $250K MAGI), asset location, direct indexing, and concentrated-stock unwind tactics.
Monte Carlo Retirement Simulation in Excel โ†’
Why the full tax cascade is one of the things Excel-based Monte Carlo builds typically can't express.
Retirement Tax Strategies (Complete Guide) โ†’
NIIT, QCDs, withdrawal sequencing, Roth conversion traps โ€” the full reference companion to this article.