Every release bumps the version shown in the calculator's header. The monthly refresh of lender terms, income-asset figures and BTC closes doesn't change the version; the header shows its "data as of" date (now 2026-10-01).
- v13.15 Monte Carlo mode: a “Monte Carlo: 1,000 random price paths” panel under the stress tests, off until “Run Monte Carlo”; each path is its own simulate of the plan as set (intraday dip, margin calls and the lender's liquidation terms exactly as in the model) with the BTC path swapped in through the
inp.stress.pathhook; paths are a circular block bootstrap of the real monthly log returns in data.json btcHistory from Jan 2014 (12-month blocks, with replacement), re-centred to mean 0 and laid on the model price path's own monthly moves from the starting price (ln P(m) = ln P(m−1) + ln(M(m)/M(m−1)) + e, M = the model path with its cycle wave; owner's call 2026-10-14), so the median path follows the model path but keeps history's volatility; seeded mulberry32 with a per-path stream (seed 1 by default, “New seed”, “Default seed”), so the same seed always gives the same numbers; shows full and partial liquidation odds and first margin calls (in all and by years 1/3/5/10/end), p10/p50/p90 of year-1 monthly income (Loans only: draws), total income, plan total, just holding and plan − holding, the share of runs beating holding, and a fan chart (p10–p90 band + p50, plan total or BTC price, with the model path; keyboard and “Show the numbers” like the other charts); runs in Web Workers built from the page's own model code (no second copy of the maths; chunked main-thread fallback), about 1 s for 1,000 × 20 years, cancelled at once by any change and rerun 400 ms after the controls settle; model mode only (a backtest says so); prints once run (a pending run is finished first); not in the link; newmonte_carlo_runevent; the page and methodology say the model's cycle added on top of history's moves makes year-to-year swings ~6–13% wider than history's, so the liquidation odds lean high (Inara QA); chart year labels space out on narrow screens; baseline maths unchanged. - v13.14 stress-test scorecard: a "Stress tests: replay real events" panel under the compare panel; "Run stress tests" runs each stress as its own simulate over a copy of the current inputs and scores it against the plan as set (outcome: survived / margin-call months / partly or fully liquidated and when; income over the whole run and the first 3 years; worst LTV; plan total, vs the plan and vs just holding; BTC left): the 2022 bear market with STRC at $71 (start price × the real closes ÷ the Nov 2021 close until BTC first closed back above it, Feb 2024, then the run's own monthly moves; income assets fall with BTC so STRC is at data.json
strc.lowat the BTC low), a 12-month dividend suspension from month 1, loss of ROC status after year 1, and a lender failure (Celsius/BlockFi style) at the lowest close in the loan's first 4 years: the loan is set off against the BTC and the rest is frozen 24 months then returned (after which the plan borrows again from a new lender on the same terms), or lost (no new loan); after the first run it follows the controls, debounced like the safe-income finder; prints once it has been run; not in the link; a stress that comes out ahead on income or plan total is shown dimmed with the reason, never green as a "better" result (Inara QA); newstress_runevent; the 2022 path, the replayed income-asset shock and the lender failure go through a newinp.stressinput that only these runs set, so baseline maths unchanged. - v13.13 link restore: a link or saved scenario with a build phase or dividend-suspension start beyond 20 years (possible on a horizon over 20 years), or a monthly income above $30,000 set via "Use $X/mo", now restores as saved instead of being cut to the slider's starting limit (20 years / $30,000); no maths change.
- v13.12 compare two scenarios: "Pin this as A" in a "Compare two scenarios" panel under the tiles keeps the current scenario as A, and whatever the controls then hold is B; the income, BTC price, LTV and wealth charts draw A's main line (income to you, BTC price, LTV, plan total) dashed and lighter behind B's, with "A: …" / "B: …" legend entries, and the tooltip, keyboard inspection and "Show the numbers" tables carry both; a table sets A, B and B − A side by side (income in year 1, total income, liquidation price and the BTC drop that triggers it, peak LTV, margin-call months, liquidated, plan total, plan total vs just holding, BTC held at the end), green when B is better, both in the controls' dollar setting; each scenario gets a short label (lender · LTV or monthly draws · strategy, plus the backtest start) and the panel names the controls that differ; "Edit A" loads A into the controls (B stays pinned, and the button becomes "Edit B"), "Swap A ↔ B" exchanges them, "Clear comparison" keeps the controls' scenario; the pinned scenario is in the link (
cmpA=holds A's encoded scenario while the controls are B,cmpB=while they are A) and in this browser's saved scenario, so a compare link reproduces both and links without it are unchanged; A is simulated only when it changes, so dragging a slider costs one run as before; the printed summary names both and prints the table; the CSV stays B only; newcompareevent; no maths change. - v13.11 contrast: the primary button and the selected segmented-toggle button use a darker orange fill so their white text passes WCAG AA (4.6:1 in light and dark; was 3.2:1 / 3.9:1), and the light theme's muted hint text is darker (#6b6a65, 4.9:1 on the page background; was 3.9:1), also on the methodology, changelog and 404 pages; chart lines, series colours and dark-theme text are unchanged; no maths change.
- v13.10 guided first run: three questions replace the start-price dialog: how much BTC (with the starting price and its live-price fetch), the monthly income wanted (whatever the dividends pay, or a set amount: the "Monthly income I want" target, strategy unchanged) and risk (Cautious / Balanced / Bold = starting LTV 15 / 25 / 30%, inside the default lender's stated 10–30% starting range, and capped at its max LTV as a safety net); they set up a fresh scenario and then show the results. Skipped for a scenario restored from a link or this browser, as before; × / Escape / Skip keep the defaults; "Start over: 3 quick questions" in the sidebar reruns it (defaults + the answers; closing it keeps the current scenario). The sidebar shows the essentials (BTC, starting price, cycle starting point, horizon, strategy, monthly income, pause-draws limit, lender, LTV) and folds the rest into "Advanced settings": remembered open or closed per browser, opened by an input error inside it, with a note of how many settings differ from the defaults (printing still lists them all).
scenario_runfor this browser's saved scenario now reportssaved(v13.4–v13.8 sentlink); no maths change. - v13.9 theme toggle: the look is unchanged and still follows the device's light or dark setting by default (Auto); an Auto / Light / Dark switch in the header overrides it, remembered in this browser (
bliThemein localStorage) and also applied to the methodology, changelog and 404 pages; printing stays light; no maths change. - v13.8 export: CSV year table and a printable summary; no maths change.
- v13.7 mobile summary bar: in the one-column layout (980 px and narrower) a bar fixed at the top of the screen shows the verdict's short forms (year-1 income, the BTC drop to liquidation or when it happened, ahead of or behind holding) in the verdict's tone colours, updating live while the controls are edited; it hides while the results are in view, sits at the top so the on-screen keyboard never covers it, reserves its own height (plus the safe-area inset) and keeps a focused input clear of it; tapping it jumps to the "In short" box and moves focus there; it is a real button whose accessible name reads the full verdict lines, not a live region; it replaces v12.4's floating "See results" button; no maths change.
- v13.6 verdict line: an "In short" box above the tiles gives the headline answer in three lines: year-1 monthly income (or why nothing reaches you in year 1: build phase, later loan start, interest and tax), the BTC drop from the starting price that liquidates at the starting LTV (partly, for a partial-liquidation lender; in a backtest, from the start month's close; when the run is liquidated it says when instead; with no starting loan, the tightest month), and the plan total ahead of or behind just holding at the end; built from the run's own figures by a standalone
buildVerdictfor reuse; no maths change. - v13.5 accessibility: every sidebar label is tied to its control and every input has an accessible name (incl. the asset-mix rows); the segmented toggles (interest, income-asset risk, price-path preset) are labelled groups with aria-pressed kept in step, also on restore from a link; charts are focusable (arrow keys, Page Up/Down, Home/End step the inspected month, announced through a live region), tap to inspect on touch, and each has a "Show the numbers" year-end table; the journey year chips are buttons (Tab, arrow keys, Enter/Space); the tooltip shows "–" instead of "NaN%" for months with no value; no maths change.
- v13.4 launch gates: the start dialog is skipped when a scenario is restored (link or this browser's last scenario) and has a close button and Escape; the safe monthly income search is debounced (results stay live while dragging, the safe figure follows ~250 ms after input settles); "How the model works" rewritten against the code, plus a methodology page and this changelog; STRC hint dates and the header's "data as of" come from data.json; the "income is small" note says nothing reaches you in year 1 instead of a negative figure; a real 404 page. No maths change.
- v13.3 the intraday dip is checked before the month-end margin-call cure: a fast crash can liquidate before the cure lands; the dotted "LTV at intraday dip" line now shows the uncured LTV. With dip 0% results match v13.2, except after a full liquidation that leaves a shortfall: v13.2 kept re-liquidating every month and added the shortfall up again, and v13.3 stops (the correct behaviour). The liquidation alert's suggestions read "a, b or c" (the defence-off backtest alert had lost its "or").
- v13.2 analytics loads through a Google Tag Manager container (
GTM_ID) instead of gtag.js, as on the CFC site; no maths change. - v13.1 Google Analytics on the public domain only (off until
GA_IDis set), privacy note; no maths change. - v13.0 risk model: liquidation also checked at an intraday dip below the close in down months (default 20%, sold at the dip price); lender liquidation terms (
liqModefull/partial,liqTargetLtv,liqPenaltyPct) replace the fixed full sale + 2%; backtests size the income-asset shock against a fixed 75% "full bear" (no longer the hidden drawdown slider); STRC/SATA rate glides over a fixed 20 years ("rate in 20 years"); LTV chart draws the real target (cycle steps; draw-pause limit in Loans only); APR/fee picked from lender size tiers (tiers). With dip 0%, full liquidation and a 2% penalty, 20-year projections match v12.8 exactly. - v12.10 honest first screen: cycle start defaults to "Where BTC is today", new "Worst case: today is the top" option (phase=peak), "Best case" relabel for phase=low; live price on first visit with the latest month-end close as fallback (no more fixed $110,000); default lender Arch Perpetual Income; "tightest month" replaces the misleading "closest call"; a plain-words note on why a thin-spread result is small
- v12.9 input checks: BTC and starting price must be above 0, at least one IPA asset above 0%; bad input shows an inline error and keeps the last valid result; "Live price" failures show inline instead of a pop-up.
- v12.8 ROC is per-lot by default; losing Strategy ROC status is an opt-in stress test.
- v12.7 preferred tax lots by purchase year table (when each batch turns taxable)
- v12.6 "Plan total" = still in the machine + all income taken (chart, tile, year card, table)
- v12.5 "Monthly income I want" in both strategies (IPA: target from dividends, excess reinvested, optional borrowed top-ups)
- v12.4 strategies renamed "Loans to other IPA" / "Loans only", UI pass (income-stops alert, mobile results button)
- v12.3 add your own cash every month (BTC and/or income assets)
- v12.2 cycle badges are traffic lights (green Y1, yellow Y2/Y3, red Y4)
- v12.1 history fixed: month M close = blockchain.info point dated the 1st of M+1 (the series is stamped one day late); backtest start clamped; strip/band scroll together; loan-start wording
- v12.0 backtest mode (real month-end BTC closes from data.json btcHistory), live-off-the-loan always projects from today with a later loan start, 4-year-cycle badges on the journey strip, draws card shows today's and nominal $
- v11.2 interest button highlight fix in spend mode
- v11.1 QA nits: draw room measured after interest, reserve→accrue in spend, liquidation tax kept out of spend income, wording/labels, $10 draw steps
- v11.0 live-off-the-loan rebuilt: choose monthly income (optionally inflation-adjusted), borrowed monthly with a pause-draws LTV cap, "safe monthly income" finder, asset settings/columns hidden in that mode
- v10.1 QA nits: fee slider 0.01 steps, spend-mode alert/subtitle, hide zero interest row
- v10.0 loan start month + loan term; renew or refinance at each maturity (not January); renewal fee on whole/extra/none; "live off the loan" mode; Arch standard + Perpetual Income presets from Arch's published terms (2026-09-29)
- v9.1 QA display nits: no refinance rows after liquidation, "+$X borrowed", start vs new borrowing split, no duplicate re-borrowed pill
- v9.0 cycle-year mode now refinances each January (new loan pays off the old; cash-out buys assets; never sells assets to pay down; rolls over if BTC hasn't risen), with refinance fee on whole new loan or extra only
- v8.4 Year 0 (start) row, start block + card, target-LTV column in the year table
- v8.3 "Today is the cycle low (live price)" option (phase=low)
- v8.2 pick the starting cycle year (Y1–Y4) next to the starting price (phase=y1..y4; old phase=peak links still work)
- v8.1 year card splits margin-call vs cycle pay-down sales
- v8.0 cycle-year LTV targets (Y1–Y4, January rebalance); margin-call cures now aim at min(target, margin-call LTV − 5 pts) in all modes (deliberate change from v7.2)
- v7.2 badge no-wrap on phones
- v7.1 QA re-check nits: card interest row, alert order, ROC tile during suspension, mobile width
- v7.0 QA fixes: drawdown realised as set, fixed 20-yr CAGR glide, peak LTV before cure, income tile breakdown, SATA ROC, fixed coupons, annual accrue, short-term gains, after-tax loan cover, liquidation shortfall, crash-at-start stress
- v6.1 version badge
- v6 multi-asset income mix
- v5 per-lot ROC tax
- v4 10–50 yr horizon, year detail card
- v3 build→payout phases, new look
- v2 inflation/STRC risk/taxes
- v1 base calculator