Skip to content

Historical backtest

How did a four-sleeve portfolio move through history?

— views

Loading counts…

Explore a draft price-history illustration from May 2007 through September 2026. It uses four ETF proxies requested from Twelve Data with adjust=all. The provider’s materials do not confirm that this series represents distributions reinvested, so the figures are not presented as verified total returns.

← Back to The Permanent Portfolio

The model being illustrated

One proxy for each of the four sleeves.

These ETFs stand in for the four asset types. The chart starts in May 2007, the first month all four have data, so it shows only the period covered by those funds. It calculates a hypothetical result from past prices; it does not show the returns any particular investor earned.

  1. 01U.S. stocks · IVV25%
  2. 02Long-term U.S. government bonds · TLT25%
  3. 03Treasury bills · BIL25%
  4. 04Gold · GLD25%

These model weights are not a suggested allocation for you. Read the Permanent Portfolio guide for an explanation of each sleeve.

Evolution · May 2007 to September 2026

How the standard and customized portfolios moved.

Both lines begin with a hypothetical $10,000 portfolio. The graph shows how that dollar value changes over time. The standard portfolio holds 25% in each sleeve and resets to those weights each December. Configure the customized portfolio’s weights and rebalancing rule to compare its path.

Evolution allocation: IVV 25% · TLT 25% · BIL 25% · GLD 25%

The standard comparison resets to 25% per sleeve each December. The sliders and drift check settings change only the customized line in this Evolution graph; summary, annual and monthly returns, and inflation below remain based on the standard portfolio.

Evolution of the standard and customized portfolio allocationsMonthly month-end hypothetical portfolio values in U.S. dollars from May 2007 to September 2026. Both lines begin at a hypothetical $10,000. The standard portfolio holds 25% in each sleeve and resets to those weights each December. The customized allocation line reflects the chosen weights and rebalancing rule. The path compounds the closing-price series returned for requests with adjust=all. It does not separately collect or reinvest cash distributions and is not a verified total-return series.$0$10,000$20,000$30,000$40,0002007-052012-012017-012022-012026-09Hypothetical portfolio value (USD)

Hover or tap a point or bar to inspect its value. For keyboard access, focus a line chart and use the arrow keys; focus a bar to inspect it.

Data source: Twelve Data closing prices requested with adjust=all, retrieved October 27, 2026.

End-of-period comparison

Each portfolio starts with $10,000 on May 31, 2007. Values are through September 30, 2026.

Standard portfolio

IVV 25% · TLT 25% · BIL 25% · GLD 25% · Annual reset · December

Ending value
$34,550
Cumulative adjusted-price change
245.5%
Annualized adjusted-price change
6.6%
Largest month-end drawdown
-15.7%

Customized allocation

IVV 25% · TLT 25% · BIL 25% · GLD 25%

Ending value
$34,550
Cumulative adjusted-price change
245.5%
Annualized adjusted-price change
6.6%
Largest month-end drawdown
-15.7%
  • Standard portfolio 25% each
  • Customized allocation
Monthly month-end hypothetical portfolio values in U.S. dollars, calculated from the returned adjusted-price series. The standard portfolio resets to 25% per sleeve each December; the customized line uses the chosen reset rule or drift-check cycle. Dividend treatment is not verified as reinvested total return.
View sampled ETF proxy index values
Each proxy’s first adjusted close = 100; these are indexed values, not dollar prices
DateU.S. stocksLong-term U.S. government bondsTreasury billsGold
May 2007100.0100.0100.0100.0
Dec 2012105.1143.6104.2247.2
Dec 2017218.4171.7104.8188.7
Dec 2022341.4149.4110.6258.8
Sep 2026715.6135.3130.4581.1

Summary · adjusted-price illustration

What happened to a hypothetical $10,000?

The same starting amount and rule are used for every measure below. Values compound the closing-price series returned for adjust=all; they are changes in that series, not verified total returns with distributions reinvested.

Starting value
$10,000

On May 31, 2007

Ending value
$34,550

On Sep 30, 2026

Cumulative change from adjusted prices
245.5%

Over 19.3 years

Annualized change from adjusted prices
6.6%

Compounded rate; not verified total return

Largest month-end drawdown
-15.7%

Peak to trough

Ending value in start-date dollars
$21,323

Adjusted with U.S. CPI-U

The annualized figure is a compounded rate calculated from the beginning and ending values of the returned adjusted-price series; it does not describe a typical year or verify distribution reinvestment. Drawdown is measured from monthly closing values, so an intra-month decline may be larger.

Price changes · provider-adjusted closes

Adjusted-price changes varied by year and month.

Annual bars show calendar-year changes in the returned adjusted-price series. The monthly histogram groups each month-end change into a fixed range. These are not verified total returns.

Data source: Twelve Data closing prices requested with adjust=all, retrieved October 27, 2026.

Annual price changes

Annual portfolio adjusted-price changesCalendar-year changes in the returned adjusted-price series for 20 years. The first period begins May 2007 and the last ends September 2026; these two periods are partial.-25%-12%0%12%25%2007 · partial period: +8.34% change in the adjusted-price series07*2008: -0.56% change in the adjusted-price series082009: +6.55% change in the adjusted-price series092010: +12.25% change in the adjusted-price series102011: +10.40% change in the adjusted-price series112012: +5.85% change in the adjusted-price series122013: -2.38% change in the adjusted-price series132014: +9.65% change in the adjusted-price series142015: -2.83% change in the adjusted-price series152016: +5.36% change in the adjusted-price series162017: +11.10% change in the adjusted-price series172018: -1.58% change in the adjusted-price series182019: +16.31% change in the adjusted-price series192020: +15.44% change in the adjusted-price series202021: +4.98% change in the adjusted-price series212022: -12.20% change in the adjusted-price series222023: +11.67% change in the adjusted-price series232024: +12.17% change in the adjusted-price series242025: +22.47% change in the adjusted-price series252026 · partial period: +0.86% change in the adjusted-price series26*

Hover or tap a point or bar to inspect its value. For keyboard access, focus a line chart and use the arrow keys; focus a bar to inspect it.

*The first period starts in May 2007; the last ends in September 2026. Both are partial years. The percentages show changes in the returned adjusted-price series, before fees and taxes.
View annual price-change values
Annual change in the returned adjusted-price series
YearPrice change
2007*+8.34%
2008-0.56%
2009+6.55%
2010+12.25%
2011+10.40%
2012+5.85%
2013-2.38%
2014+9.65%
2015-2.83%
2016+5.36%
2017+11.10%
2018-1.58%
2019+16.31%
2020+15.44%
2021+4.98%
2022-12.20%
2023+11.67%
2024+12.17%
2025+22.47%
2026*+0.86%

Monthly price-change histogram

Distribution of monthly adjusted-price changesHistogram of 232 month-to-month changes in the returned adjusted-price series. It groups observations into seven fixed percentage intervals; dividend treatment is not verified as total return.0224466≤ −5%: 2 months≤ −5%−5% to −3%: 7 months−5% to −3%−3% to −1%: 40 months−3% to −1%−1% to +1%: 87 months−1% to +1%+1% to +3%: 69 months+1% to +3%+3% to +5%: 20 months+3% to +5%≥ +5%: 7 months≥ +5%

Hover or tap a point or bar to inspect its value. For keyboard access, focus a line chart and use the arrow keys; focus a bar to inspect it.

Each observation compares one month-end portfolio value with the previous month-end using the returned adjusted-price series. It does not separately add or reinvest cash distributions.
View monthly return counts
Count of month-to-month adjusted-price changes by range
Monthly price-change rangeMonths
≤ −5%2
−5% to −3%7
−3% to −1%40
−1% to +1%87
+1% to +3%69
+3% to +5%20
≥ +5%7

Risk · individual proxy histories

How did the standard portfolio compare with the stock and gold proxies?

The standard line is the hypothetical portfolio with 25% in each of IVV, TLT, BIL, and GLD, reset to those weights each December. The other lines show IVV and GLD individually. A larger historical price gain alone does not tell you how deep a decline may be or how long recovery may take. These are the backtest’s proxies, not every ETF listed for those buckets.

Decline from each proxy’s previous month-end high

Month-end drawdowns for the standard portfolio, IVV stock proxy, and GLD gold proxyMonthly month-end drawdowns for the standard 25%-each portfolio, IVV U.S. stock proxy, and GLD gold proxy, from May 2007 through September 2026. Each line measures its decline from its own prior month-end high. The proxy series have different historical peaks and troughs.0%−20%−40%−60%200720112015201920232026Decline from previous month-end high

Hover or tap a point or bar to inspect its value. For keyboard access, focus a line chart and use the arrow keys; focus a bar to inspect it.

  • Standard portfolio 25% each · December reset
  • U.S. stock proxy · IVV
  • Gold proxy · GLD

Data source: Twelve Data month-end closing prices requested with adjust=all, retrieved October 27, 2026.

Each decline is measured against that proxy’s own previous high. A recovery is the first month-end close at or above that high. The stock and gold proxies reached their peaks at different times, so the two declines do not represent a simultaneous portfolio loss. Month-end observations can miss larger intra-month declines.

Standard portfolio · 25% each · December reset

-15.7%

Peak Dec 2021 · trough Sep 2022

Decline from peak to trough: 9 months.

Recovery to prior high: 18 months after trough · regained Mar 2024. Below prior high for 27 months in total.

U.S. stocks · IVV

-50.8%

Peak Oct 2007 · trough Feb 2009

Decline from peak to trough: 16 months.

Recovery to prior high: 37 months after trough · regained Mar 2012. Below prior high for 53 months in total.

Gold · GLD

-42.9%

Peak Aug 2011 · trough Dec 2015

Decline from peak to trough: 52 months.

Recovery to prior high: 55 months after trough · regained Jul 2020. Below prior high for 107 months in total.

Impact of inflation

A dollar amount and its purchasing power are different.

The solid line shows the portfolio value before inflation adjustment. The dashed line shows its purchasing power in May 2007 dollars, adjusted using U.S. CPI-U.

Nominal and inflation-adjusted portfolio valueA $10,000 starting value ends at $34,550 nominally and $21,323 in May 2007 dollars, using monthly U.S. CPI-U all-items values.$0$10,000$20,000$30,000$40,0002007-052012-012017-012022-012026-09

Hover or tap a point or bar to inspect its value. For keyboard access, focus a line chart and use the arrow keys; focus a bar to inspect it.

  • Portfolio value before inflation adjustment
  • Purchasing power in May 2007 dollars (CPI-adjusted)
Inflation is measured with the U.S. CPI-U all-items index, not seasonally adjusted. It is a broad U.S. price measure and may not match any household’s costs or Canadian inflation. CPI data retrieved October 26, 2026. The BLS series has no all-items value for October 2025; the line leaves that month blank and does not estimate it.
Portfolio value before inflation adjustment
$34,550

On Sep 30, 2026

Purchasing power after inflation
$21,323

In May 2007 dollars · CPI-adjusted

Cumulative U.S. CPI-U inflation
62.0%

May 2007 to Sep 2026

Over this interval the CPI-U index rose 62.0%. The inflation-adjusted series divides each portfolio value by the CPI change since the first observation.

How to read this calculation

The displayed history has a defined scope.

  1. Market data. Daily closes for IVV, TLT, BIL, and GLD were requested from Twelve Data with adjust=all. The API lists all, splits, dividends, and none as adjustment modes; Twelve Data’s support FAQ describes historical prices as split-adjusted and says additional adjustment may require client-side tools. The charts use each month’s last shared trading-day close and stop at September 30, 2026, the latest month with a matching CPI observation.
  2. Portfolio rule. The starting value is $10,000, split equally among the four proxies at the first common month-end. The portfolio is rebalanced back to 25% per proxy at each December month-end.
  3. Return treatment. The calculation compounds changes in the returned adjusted-price series. It does not separately collect or reinvest cash distributions, and provider documentation does not confirm that these observations equal a dividend-reinvested total-return series. These figures are therefore not verified total returns. Fees, taxes, and transaction costs are also excluded.
  4. Inflation. The comparison uses monthly U.S. CPI-U, all items, not seasonally adjusted. The inflation-adjusted portfolio values are expressed in May 2007 U.S. dollars. This broad U.S. measure may not reflect a household’s costs or Canadian inflation.
  5. Missing CPI observation. BLS has no all-items CPI value for October 2025. The inflation-adjusted chart leaves that month blank and does not estimate it.

For education only. This hypothetical price-history illustration is not a verified total-return backtest, investment advice, a recommendation, or a forecast. Historical results do not guarantee future results.

Data and further reading

References

  1. Curvo Backtest — a presentation reference only. This page uses its own selected proxies, source observations, and calculations.
  2. Twelve Data API documentation — lists the time-series adjustment modes, including all, splits, dividends, and none.
  3. Twelve Data adjustment FAQ — describes historical prices as split-adjusted and notes additional client-side adjustment tools.
  4. Twelve Data commercial-use guidance and EOD pricing guidance — explain that public display depends on plan tier and applicable exchange licensing.
  5. Twelve Data attribution guidelines — require a dofollow link and applicable exchange-specific attribution.
  6. U.S. Bureau of Labor Statistics CPI series IDs — identifies CPI-U, U.S. city average, all items, not seasonally adjusted.
  7. BLS note about the missing October 2025 CPI observation — explains the gap shown in the inflation-adjusted chart.
  8. Investor.gov guide to asset allocation, diversification, and rebalancing — general educational context.
  9. Investor.gov guide to risk tolerance — discusses how the time horizon for a goal relates to investment risk.

← Back to The Permanent PortfolioAll learning topics