Retired, 66 portfolio 30

user Retiree at 66 · defined 2026-10-02 18:49 · an example: a common US account mix in retirement (made-up figures); optimized for income · every portfolio

Optimized for income

made 2026-10-02 18:50 by 4_book.py · objective: largest withdrawal at the book's ruin, over 30 years, raised 2% a year, at 5% risk of ruin · 2003-01-02 .. 2026-09-25

Try your own version of this portfolio →add, remove and re-weight its strategies; worked out in your browser, nothing stored

Have a portfolio like this built for your own accounts

The portfolios on this site are pre-tax and use made-up accounts — they cannot know which accounts you hold (RRSP, TFSA, non-registered; or 401k, IRA), what is in each, or the tax drag that applies to your situation. A portfolio built for your own money needs all of that. That is why it runs on your computer, not here.

With Cupel Portfolios

Cupel Portfolios models your account types and the tax drag in each, and runs the same ratings and the same builder as these examples — fitted to your actual balances, optimized to settings you choose (growth or income, the worst fall you will accept). It runs on your own computer; your account details and balances never leave it. Download Cupel Portfolios free. Sign up for a Cupel account to get fresh data through the API.

Which runs passed

23 candidates · 5 held

primary: every usable DONE run → grade_min B: 29 → exclude 870: 28

23 candidates after the rest of the screens (crypto, the same curve twice, shorting, the window, the quick build's); 5 held · probability of backtest overfitting over the candidates 40.85%

Figures

2003-01-02 .. 2026-09-25

withdrawal 52,773/yr at 5.00% ruininvested 700,000 on 700,000 of capital (no margin)primary: 5 strategies, 100.00% of the portfolio

CAGR

16.83%

2003-01-02 → 2026-09-25 (23.7 years)

Sortino

1.383

daily, target 0

Max drawdown

45.09%

2007-10-31 → 2008-11-20, recovered 2009-11-25

Total return

3911.42%

700,000 → 28,079,937 undefined

Sharpe

0.955

risk-free 0

Volatility

17.99%

annualized, daily returns

Calmar

0.373

CAGR / max drawdown

Longest under water

756 days

2007-10-31 → 2009-11-25

Best year

50.39%

2009

Worst year

-31.62%

2008

Positive months

65.26%

of 285

2003-01-02 700,000 → 2026-09-25 28,079,937 · log scale · 6085 days · hover or use the arrow keys for values

Drawdown: the fall below the highest value reached so far. Worst 45.09% below its peak, from 2007-10-31 to 2008-11-20, back to that peak 2009-11-25 (756 days from the peak) · 8.67% below its peak on the last day

30-year Monte Carlo

the withdrawal and the spread of wealth

Safe withdrawal, 5.0% ruin

52,772/yr

7.5% of 700,000, raised 2.0% a year; ruin 5.0%

Same, never raised

61,496/yr

8.8% of 700,000, flat; ruin 5.0%

Median wealth, year 30

33,984,025

paying the raised withdrawal

5th percentile, year 30

2,121

paying the raised withdrawal

Ruin with no withdrawal

0.0%

paths that reach zero on returns alone

History sampled

284 months

2003-02 → 2026-09

By year, log scale (under 10,000, ruin included, sits on the floor) · line median, dashed 5th percentile, bands 25th–75th and 5th–95th · hover or use the arrow keys for values

The chart is wealth while paying 52,772 a year raised 2.0% yearly. 2,000 paths of 30 years, each built from 12-month blocks of the portfolio's own monthly returns (2003-02 → 2026-09), drawn at random. A withdrawal is taken at the start of each month; ruin is a path that cannot pay one in full. The 2.0% inflation is a guess (the Bank of Canada's target), not a measured rate. 30 years are drawn from 23.7 years of history, so the same years recur within a path. The portfolio's weights were fitted to this same history, so every figure here leans optimistic.

Walk-forward

the years the weights never saw

The whole selection (ratings, admission and the fit) rerun on each training window's data alone, then frozen and scored on the years after it, beside equal weights over that window's candidates. Reconstruction: only variations registered before the split. Retrospective: every variation, a test of the choosing procedure, not evidence those variations existed then. The held-out figure is the one to compare portfolios by.

Mode	Fitted On	Withdrawal	CAGR	Max DD	Held Out	Withdrawal	CAGR	Max DD	Equal Weights: Withdrawal	CAGR	Max DD
reconstruction	2003-01-02 .. 2012-12-31	skipped: no variation was registered before 2013-01-01 (the first was registered 2026-09-27)
retrospective	2003-01-02 .. 2012-12-31	7.65%	13.17%	21.03%	2013-01-02 .. 2019-12-31	3.84%	4.88%	11.15%	4.10%	5.87%	16.48%
reconstruction	2003-01-02 .. 2019-12-31	skipped: no variation was registered before 2020-01-01 (the first was registered 2026-09-27)
retrospective	2003-01-02 .. 2019-12-31	6.62%	15.22%	48.60%	2020-01-02 .. 2026-09-25	8.06%	18.10%	34.16%	8.32%	18.83%	34.07%

Withdrawal: a year, as a share of the start, from that window's own monthly returns.

Strategies

6 rows

Weight and amount are the portfolio's; tier is primary (drives the portfolio) or secondary (added around them); every other column is the held run as this portfolio copied it, measured in that run's currency over its whole curve. Each strategy is shown at the version the run used. "family X" marks holdings that are variations of each other in one currency (only portfolio settings differ): 1 such group.

Strategy	Market	Tier	Weight ▼	Amount	Run	Status	Grade	CAGR	Sortino	Max DD	Total Return	Sharpe	Volatility	Calmar	Under Water Days	Best Year	Worst Year	Positive Months	Fills	Fees
clear	all CADUSD	all primarysecondary				all DONE	all ABCD–
Thesis: An even split of shares put back to even once a year family A v1	USD	primary	0.2568	179,777	9496	DONE	B	19.88%	1.225	55.34%	7282.31%	0.847	25.12%	0.359	887	76.55% (2023)	-41.82% (2008)	61.75%	1,233	70,577
Thesis: There is a best gap between one evening up and the next family A v1	USD	primary	0.2061	144,248	9532	DONE	B	20.67%	1.203	65.11%	8529.23%	0.834	26.88%	0.317	1,160	70.30% (2009)	-49.92% (2008)	63.16%	8,415	343,640
Thesis: An even split of shares put back to even once a year family A v1	USD	primary	0.0696	48,690	9496	DONE	B	19.88%	1.225	55.34%	7282.31%	0.847	25.12%	0.359	887	76.55% (2023)	-41.82% (2008)	61.75%	1,233	70,577
Thesis: There is a best gap between one evening up and the next family A v1	USD	primary	0.0558	39,067	9532	DONE	B	20.67%	1.203	65.11%	8529.23%	0.834	26.88%	0.317	1,160	70.30% (2009)	-49.92% (2008)	63.16%	8,415	343,640
Thesis: An even split of shares put back to even once a year family A v1	USD	primary	0.0482	33,708	9496	DONE	B	19.88%	1.225	55.34%	7282.31%	0.847	25.12%	0.359	887	76.55% (2023)	-41.82% (2008)	61.75%	1,233	70,577
Thesis: There is a best gap between one evening up and the next family A v1	USD	primary	0.0386	27,047	9532	DONE	B	20.67%	1.203	65.11%	8529.23%	0.834	26.88%	0.317	1,160	70.30% (2009)	-49.92% (2008)	63.16%	8,415	343,640

Contribution

what each strategy adds

Rating 6 (STRATEGY-RATINGS.md §5.2): what each strategy adds to this portfolio's withdrawal — the portfolio re-optimized without it, in every account it is held in. It belongs to this portfolio and means nothing outside it. Raw: the curves as they ran. Conservative: the portfolio fitted again with 25% of each run's alpha over its index kept; a strategy only that fit holds is a near-miss. A strategy a fit does not hold adds 0 to it. The band is the 90% range over paired resamples of the simulated paths: a difference inside it is not a ranking. Cluster: the strategies whose daily returns correlate 0.9 or more with it (it included), removed together.

Strategy	Market	Held	Raw [90% Band]	Conservative [90% Band]	Cluster	Cluster Removed: Raw	Conservative
clear		all heldnear-miss
Gold and long Treasuries, risk parity at a 10% volatility target	USD	held	+825 [+90 .. +1,405]	+829 [+226 .. +1,805]	1	+825	+829
Thesis: An even split of shares put back to even once a year	USD	held	+822 [-137 .. +2,132]	+2,590 [+1,506 .. +3,280]	4	+13,938	+7,323
Thesis: There is a best gap between one evening up and the next	USD	held	+505 [-441 .. +1,438]	+0	4	+13,938	+7,323
Gold trend timing: gold above its 10-month average, else T-bills	USD	held	+15 [-648 .. +310]	+0	1	+15	+0
OSAP Coskewness / skewness (low)	USD	held	-120 [-837 .. +425]	+1,769 [+917 .. +2,643]	1	-120	+1,769
Thesis: Putting a wider list back to even earns more	USD	near-miss	+0	-1 [-237 .. +195]	4	+13,938	+7,323

Report a problem or ask a question about this portfolio.