For more details, consult the README.md.

• 11.129d Annual Details now shows Roth conversions as the IRA withdrawals they are, and conversions draw from the larger IRA. A Roth conversion is money leaving the IRA, so the IRA WD, IRA1- and IRA2- columns now include it. Before, a year could show a large conversion with little or no IRA withdrawal, which looked wrong even though the balances were correct. RMD stays in its own columns (it is the involuntary draw and is never a conversion). The Federal and State tax columns now include the tax on conversions, so they add up to Total Tax and the Taxation chart shows the real tax. Conversions are now taken from the larger IRA first (spilling to the smaller only when the larger cannot supply the full amount) instead of being split by size, which is what you would actually do and avoids converting a token slice out of a tiny IRA. This changes each IRA's balance over time and therefore each person's future RMDs; the combined yearly tax is unchanged. Clicking a year to open the Tax Payment Planner now sends the correct per-IRA voluntary, conversion, and tax figures.
• 11.1287 Maximize Conversions now actually maximizes: available Cash can cover conversion taxes. Previously nothing in the tool paid a conversion's tax bill from outside funds, so a $20,000 Extra Annual Roth Conversion only landed about $13,700 in Roth - the rest went to the tax on its own withdrawal. Standard practice is to cover that tax from Cash so the full amount gets to grow tax-free, and Maximize Conversions now does exactly that, on top of its original behavior of routing leftover IRA withdrawals into Roth. It uses only the Cash you have: when Cash is short it funds what it can, and at $0 Cash nothing changes, so existing plans are unaffected until you turn it on. Turn on nerd knobs to control the two behaviors separately ("Convert Excess to Roth" and "Use Cash"), in which case the Optimizer also sweeps each strategy both with and without cash-funded taxes (💵 rows).
• 11.1287 Optimize Spend and Optimize Conversions moved to the Optimizer tab, and Optimize Conversions is now on by default. Both only ever affected the Optimizer, but sat in the sidebar next to the strategy settings that drive every tab, which made them look like plan settings. They now live at the top of the Optimizer tab under "Search options", where they apply. Optimize Conversions starts on so the conversion-optimized ⇌ rows are found without having to know to ask for them; turn it off if you want a faster sweep. Note that ⇌ rows only appear where an extra conversion actually improves the result for one of the top-ranked strategies, so a plan whose best strategies do not benefit from converting more will correctly show none. Your saved scenarios and shared links keep working unchanged.
• 11.1287 Smaller fixes. The Roth Conv column in Annual Details had silently lost its explanatory tooltip to a naming typo (as had the Roth Growth column); Roth Conv's now spells out why it can read lower than the amount you asked to convert. The Break Even explainer is a compact ⓘ next to the year instead of a standing button, and it no longer makes you click before telling you anything - hover reads the reason, click pins it open, click again closes it.
• 11.1271 Three fixes from testing the Extra Conversion field and Break Even. (1) The Optimizer's conversion-amount search could recommend a Guyton-Klinger conversion so large that GK could only "afford" it by continuously cutting your future spend behind the scenes -- the search now rejects any amount that isn't actually sustainable for GK, the same stability check already used elsewhere in the optimizer. (2) The "Extra Annual Roth Conversion $" tooltip now says plainly that it is capped only by the remaining IRA balance, not by your IRA Goal -- a conversion moves money IRA-to-Roth rather than out of the household, so it is allowed to draw the IRA below that goal on purpose. (3) When Break Even shows “—”, a new ⓘ next to the stat identifies the specific conversion year that erases an otherwise-sustained lead, instead of leaving you to guess why.
• 11.1253 Optimizer strategies loaded from the table now actually match what the table showed. Clicking a "⇌ Optimize Conversions" row to load it previously dropped the extra conversion amount that made that row special, so the loaded plan's Break Even (and everything else) could disagree with the Optimizer table. A new "Extra Annual Roth Conversion $" field now holds that amount, is filled in automatically when you load a ⇌ row, and is shareable/saveable like any other input. Loading a cyclic-brokerage or IRMAA/ACA-ceiling ⇌ row now also correctly carries those settings over. Also clarified the Conv Savings tooltip: it only counts tax actually paid so far, so it can look good even when Break Even (which also counts the tax still owed on money left in the IRA) says the conversions never really paid off.
• 11.1247 Optimizer: Break Even now shown for Optimize Conversions strategies. When the Optimize Conversions checkbox is on, the top 5 strategy rows now report a Break Even year (same permanent-crossover definition as the single-scenario stat) in a new column, and a new "Earliest Break Even" objective (nerd mode) ranks strategies by how soon their conversions permanently pull ahead. Strategies that never sustain a lead show “—”.
• 11.1240 Break Even now requires the lead to hold for the rest of the plan. A rare scenario could report Break Even off a single year where the converting (or excess-withdrawal) plan brushed even before falling behind again, sometimes for good. Break Even is now the earliest year the plan pulls ahead and stays ahead through the end of the simulation, showing “—” if that lead is never sustained. Same fix applies to excessOC. See “What is Break Even?” below.
• 11.11ff Refactor: No behavior changes. If the page looks stale after this update, do a hard refresh.
• 11.11f3 Internal cleanup: source files renamed to a shorter, consistent scheme (optimizer_core.js, optimizer_ui.js, optimizer_tests.js, optimizer_text.js, optimizer_styles_responsive.css). The simulation engine and the page UI now live in separate files. No behavior changes. If the page looks broken after this update, do a hard refresh (Ctrl+Shift+R) to clear the old cached files.
• 11.11dc Break Even and Opp. Cost are now measured with a true no-conversion re-simulation. The tool re-runs your whole plan with the Roth conversions removed and compares after-tax wealth year by year. The no-conversion plan keeps the money in the IRA and pays its own larger RMD taxes and IRMAA surcharges later, so the benefit of avoiding those is now fully counted. The old approximation could report a Break Even year when there were no conversions at all, and could fail to report one for conversions that clearly paid off. Break Even now appears only when conversions actually happen. The same rework applies to the excess-withdrawal opportunity cost (excessOC). See "What is Break Even?" below.
• 11.11c8 Improved bar charts. Hover over or click legends to isolate that value. Double click to restore. Line legends are unchanged. IRMAA is now presumed to occur in years 1 and 2 (matching the future value) - a bug caused it to only show up on year 2.
  
  Stress is now performed automatically when Historical Monte Carlo is selected and applies to the current strategy. Line charts now also have the "isolate" behavior in addition to the "click to dismiss, click to reshow."
  
  Cycle Brokerage will now fill up the 0% or 15% tax bracket (whichever falls below any current limits). Previously it was only pulling enough to meet spending.
  
  Other minor UI improvements also made.
• 11.11ae Income & Expenses chart: added a note that shown incomes are after-tax (pre-tax figures are in Annual Details).
• 11.11ad Added Spending to Annual Details to focus on spending.
• 11.1133 Architectural improvements, no features changed.
• 11.1125 IRMAA now respects Medicare age.
  • Age-65 gate, per spouse. The IRMAA surcharge is charged only for spouses who are 65+ (actually on Medicare) — a couple aged 61/59 no longer pays IRMAA no matter how large a Roth conversion is. With one spouse 65+ and one younger, only the older spouse's half is charged.
  • Tier display fix. The IRMAA Tier column (and chart milestone) previously appeared one year before the surcharge was actually charged; both now use the same 2-year MAGI lookback as the dollar amount, and show -none- before age 65.
  • IRMAA Ceil strategies unlocked pre-63. Before any spouse is 63 (the earliest year income can affect a premium, given the 2-year lookback), the IRMAA-tier conversion ceiling relaxes to the top of the federal bracket containing it — no more pointlessly capped conversions in your early 60s. Same for “Lesser of IRMAA or Bracket.”
  • Medicare Parts B+D base cost now shown on charts and table for spouses 65+ — illustration only, not deducted from spendable.
  • Distinct chart colors. IRMAA is now deep pink and Medicare teal on all charts (they previously shared nearly identical pinks); the IRMAA milestone marker matches the pink.
• 11.1119 State retirement-income taxes are now much more accurate. 16 states (Alabama, Colorado, Connecticut, Georgia, Illinois, Iowa, Kentucky, Maine, Maryland, Michigan, Mississippi, New York, Ohio, Pennsylvania, Virginia, Wisconsin) now apply their real pension/IRA exemptions, caps, and credits instead of taxing all retirement income as ordinary income. Several states have limitations; limitations are listed below the state once it's selected. See, e.g., CA, MA, NC, SC, NE, AZ, and MT.
• 11.1102 State retirement-income exemptions (Illinois & Pennsylvania): IRA/401k/pension distributions are no longer taxed by states that exempt them. Illinois and Pennsylvania now correctly exempt these withdrawals (interest, dividends, and capital gains remain state-taxed). Other states are unchanged.
• 11.10ee Pension start age, optimizer symbol legend, and a Guyton-Klinger fix:
  • Pension start age. Model a pension that begins after you retire (e.g. retire at 60 while the pension starts at 65). Leave the start age at 0 to begin at retirement.
  • Optimizer symbol legend. A new legend explains the symbols shown on strategy names (✓ ✦ ▼ 🗘 🔄 ⇌ ⚠️ 🟢 🚨 ⚓).
  • Guyton-Klinger sustainable-spend fix. When no strategy can fund your spend goal, the suggested fallback spend no longer reports a Guyton-Klinger value that only “works” by cutting spending every year — and the suggested strategy is now clickable to load.
• 11.10cf Numerous user experience improvements: IRA Draw strategies now tested to 20%, After-Tax-Spend searches higher spend rates and a suggested value is offered. Other changes include tooltip clarity improvements, chart improvements, and showing Charts first by default.
• 11.10a2 Added additional charts for clarity: Taxation, Income vs Net Income, Inflows vs Outflows, Earnings vs W/D.
• 11.1099 Baseline strategy is now ranked by total economic value, not terminal wealth alone. The pinned ⚓ baseline is the best no-conversion strategy. Guyton-Klinger Optimize Spend no longer reports an unsustainable spend. Optimize Spend now applies a stability floor and accepts a higher initial spend only if the guardrails never cut real delivered spending below one guard band of the initial.
• 11.1091 >Fixed a NaN in a Monte Carlo.
Fixed a shortfall. Fill strategies will exceed the threshold when needed to meet the spend goal. However, the ACA Cliff ceiling is now its own strict strategy: it never breaches the cap. If spending can’t be met under the target the plan is flagged untenable (⚠️ in the Optimizer) rather than silently overspending.
• 11.1060 Stress mode now ranks worst decades by real inflation-adjusted Compound Annual Growth Rate (Real CAGR): real CAGR = (1 + equity) / (1 + inflation) − 1. High-inflation decades like the 1960s–70s stagflation, and the Lost Decade (1999-2010) correctly rank as worse for retirees than a pure equity crash with near-zero inflation. Depression-era crashes also had high deflation which is now clamped to a minimum of -1% (rather than -9%).
• 11.1042 Guyton-Klinger Guardrails withdrawal strategy:
  • Dynamic spending rules. The Guyton-Klinger strategy adjusts annual spending based on portfolio withdrawal rate. It skips inflation adjustments when the prior year return was negative, and cuts spending 10% when the withdrawal rate exceeds the upper guardrail; raises spending 10% when withdrawal rate falls below the lower guardrail. Supports higher initial withdrawal rates (~5–5.5%) by absorbing sequence-of-returns risk through spending flexibility.
  • Four configurable parameters: Upper/lower guardrail (default ±20% of IWR) and cut/raise percent (default 10%). Annual Details Income category shows gkSpend and gkAdj columns.
• 11.1001 Best "Baseline" is used for strategy comparison:
  • Honest baseline. The optimizer now also runs every strategy family with no Roth conversions and no cyclic brokerage maneuvering, no QCDs, and pins the strongest of those as a ⚓ BASELINE reference row at the top of the results. Every other strategy is measured against it — because "strategy A beats strategy B" only means something relative to the best you can do without conversion/brokerage tricks.