Methodology note
How the analysis works · 6 min read
How we built a Monte Carlo retirement analysis that survives an SEC examiner.
Most household-facing retirement tools show one best-guess line. A defensible plan has to show the range, and carry the SEC disclosure with every number it produces. Here's how the new analysis on this site is built.
By Dan Zimon · May 27, 2026
Most household-facing retirement tools answer the question 'how am I doing?' with a single best-guess line. A deterministic projection at one assumed return. The line looks clean. Real markets are not clean. The range of outcomes a real plan must survive is wider than that one line suggests, and the most useful planning question is not 'what is the expected number' but 'what is the worst quarter of trials, and can the household withstand it?'
The analyzer at /retirement-analysis is built around that question. It runs 1,000 simulated paths against the household's actual numbers, current ages, retirement target, accounts and balances, contribution schedules, income streams, expense lines, and reports the full distribution. Median ending balance. 10th percentile (the bad-sequence tail). 90th percentile (the favorable tail). Plan-success rate defined plainly as 'of every 100 simulated paths, X ended above zero.' No branded confidence numbers, no implied certainty.
What the engine actually models
- Federal income tax, 2025 brackets, indexed for inflation each projection year.
- Required minimum distributions per SECURE 2.0 (age 73 today, age 75 from 2033).
- IRMAA tier tracking on Medicare Part B/D thresholds, so the bracket math accounts for the surcharge cliffs at age 63.
- NY retiree state-tax add-on, with the pension and annuity exclusion applied.
- Social Security claim-age curves per SSA-published formulas. Spousal benefits coordinated.
- Pension COLA modeling (NY public formula, fixed, or none).
- Withdrawal sequencing across taxable, tax-deferred, Roth, and cash, with a tax-gross-up so the household nets the actual spending target.
- Optional long-term-care expense stress test, modeled at population-level frequency in the tail.
What the engine does NOT model, and surfaces honestly
- Net investment income tax (NIIT 3.8%).
- Alternative minimum tax (AMT).
- Capital-gains stacking on the ordinary brackets.
- States other than NY.
- Individual security selection. Monte Carlo uses a single broad-market return distribution.
- Tax-law changes beyond the 2026 TCJA sunset.
Every analysis the tool produces carries those modeled and not-modeled lists as a structured disclosure. The disclosure is not a footer. It is a field on the artifact. Enumerated, expandable, with the rule citations underneath. SEC Marketing Rule 206(4)-1(d) governs how a registered adviser presents hypothetical performance to clients. It requires reasonable basis for the projection, sufficient information for the audience to understand the assumptions, and information underlying the projection available promptly on request. The compliance card on every result panel satisfies all three.
Why this is structured the way it is
Tools that bury the disclosure under a single hidden link are technically compliant and practically misleading. The household never reads the disclosure. The number on the page looks like a fact. When the SEC IA-5780 enforcement action came down against another firm in 2022, the language pattern the Commission cited was exactly that. Performance figures presented as fact, with the methodology and assumptions out of view. Putting the disclosure inline, structured, and tied to every number is the design choice that keeps the tool honest at the surface a household actually sees.
The side-by-side comparison
A single retirement-age scenario is rarely the actual decision. The harder question is usually 'what if I retire at 62 instead of 65. What does that cost?' or 'what if I claim Social Security at 67 vs 70?' The tool runs the comparison for you. Same household, same Monte Carlo seed across all scenarios, so the only thing changing between them is the decision being tested, not Monte Carlo noise. Three scenarios side by side, with plan-success rate, median, p10, p90, and first-shortfall year for each.
The Roth conversion ladder
Between retirement and the RMD start year, most households have an unusual window where ordinary income drops sharply but RMDs have not yet started. That window is the natural home for filling unused lower-bracket space with Roth conversions. The tool computes the recommended ceiling year-by-year, capped at the 12% federal bracket line (or the IRMAA Tier 1 threshold once the household is Medicare-relevant). It is an illustrative ceiling, not a target, and the ladder closes when RMDs begin to fill the bracket on their own.
What it is not
This is an educational tool, not advice. Camba Capital is not yet registered as an investment adviser and is not currently providing advisory services. The analysis surfaces decisions worth a conversation. The conversation belongs to a fiduciary review process that is separate from the tool itself, runs after registration completes, and operates under a documented §206 duty of care. The disclosure on every result panel says exactly that.