Portfolio Archetypes Search: Finding the Sweet Spot

Question

We tested: given the durable findings about leverage, volatility decay, gold's defensive value, and TLT's failure as defensive, what portfolio composition is Sharpe-optimal across multiple regimes?

Specifically: - Is 75% UPRO + 25% UGL (2x gold) the right shape? - Does silver 1x belong? - Does BTC have enough history to warrant inclusion? - How does any of this compare to his current 9-asset portfolio?

Important framing: this is data, not advice. decisions live with the operator. We're testing a discrete set of candidate portfolios and a sensitivity sweep around the most interesting axis, then publishing the results.

Methodology

Window 1 (stress): 2000-08-30 → 2026-05-18 (25.7 years), uses synthetic LETFs and synthetic 2x metals where the real ETFs didn't exist. Covers dotcom, GFC, COVID, 2022 rate shock.
Window 2 (bull): 2012-02-02 → 2026-05-18 (14.3 years). The "easy" window for leverage.
Window 3 (BTC-era): 2014-09-17 → 2026-05-18 (11.6 years). Only portfolios with BTC.
Synthesis: Same as prior studies. (1 + leverage × daily_return − 0.91%/252).cumprod() for 2x and 3x; 1x is raw.
Rebalance: Weekly, 1bp/side turnover cost.
Starting capital: $10,000.

Caveat on the current portfolio (A) test: for this study, ZROZ was approximated as TLT (1x long Treasuries) because the duration-based synthetic from the earlier stress test added noise. As a result, the current portfolio's window starts 2002-07 (TLT inception) instead of 2000-08, skipping the worst months of the dotcom drawdown. The Sharpe 0.840 for the current portfolio in this study is NOT directly comparable to the Sharpe 0.589 in the earlier stress test — windows differ. The bull-window (2012-2026) numbers ARE comparable across all portfolios since all assets exist by 2012.

Results: Portfolio archetypes (10 candidates)

Stress window 2000-2026

Portfolio	Yrs	CAGR	MDD	Sharpe	Sortino	Calmar	$10k →
A. Current (9-asset, ZROZ≈TLT)	23.8	25.33%	-69.54%	0.840	1.148	0.364	$2,138,719
B. 75 UPRO / 25 UGL	25.7	18.27%	-87.48%	0.604	0.769	0.209	$741,763
C. 50 UPRO / 25 UGL / 25 cash	25.7	15.86%	-72.55%	0.647	0.826	0.219	$437,356
D. 60 UPRO / 20 UGL / 20 AGQ	25.7	19.97%	-81.71%	0.654	0.829	0.244	$1,069,711
E. 50 UPRO / 25 UGL / 15 SILV / 10 cash	25.7	17.77%	-73.16%	0.676	0.862	0.243	$664,283
F. 40 UPRO / 20 TQQQ / 20 UGL / 10 AGQ / 10 cash	25.7	18.32%	-83.75%	0.624	0.812	0.219	$749,430
G. 40 UPRO / 20 UGL / 20 BTC / 20 cash	11.6	35.31%	-46.16%	1.217	1.532	0.765	$338,091
H. 50 UPRO / 25 UGL / 25 TQQQ	25.7	18.12%	-90.41%	0.591	0.769	0.200	$717,193
I. 60 SSO / 30 UGL / 10 cash	25.7	15.79%	-60.73%	0.709	0.916	0.260	$430,130
J. 40 UPRO / 20 UGL / 20 cash / 20 TQQQ	25.7	16.06%	-83.26%	0.587	0.763	0.193	$456,924

Bull window 2012-2026 (apples-to-apples for all but BTC)

Portfolio	Yrs	CAGR	MDD	Sharpe	Calmar	$10k →
A. Current	14.3	29.44%	-55.96%	0.961	0.526	$396,169
B. 75 UPRO / 25 UGL	14.3	33.79%	-64.96%	0.952	0.520	$634,770
C. 50 UPRO / 25 UGL / 25 cash	14.3	24.42%	-49.00%	0.965	0.498	$225,290
D. 60 UPRO / 20 UGL / 20 AGQ	14.3	29.32%	-62.43%	0.882	0.470	$391,024
E. 50 UPRO / 25 UGL / 15 SILV / 10 cash	14.3	25.51%	-52.20%	0.939	0.489	$255,296
F. 40 UPRO / 20 TQQQ / 20 UGL / 10 AGQ / 10 cash	14.3	31.71%	-57.50%	0.964	0.551	$507,474
G. 40 UPRO / 20 UGL / 20 BTC / 20 cash	11.6	35.31%	-46.16%	1.217	0.765	$338,091
H. 50 UPRO / 25 UGL / 25 TQQQ	14.3	37.55%	-62.84%	0.990	0.597	$942,078
I. 60 SSO / 30 UGL / 10 cash	14.3	21.20%	-41.57%	0.964	0.510	$155,118
J. 40 UPRO / 20 UGL / 20 cash / 20 TQQQ	14.3	30.58%	-53.71%	0.989	0.569	$448,767

Informed portfolios (synthesizing all prior findings)

Designed deliberately around: heavy gold, 2x over 3x where possible, cash buffer, no silver, no TLT/ZROZ.

Portfolio	Stress Sharpe	Stress MDD	Bull Sharpe	Bull MDD	Stress $10k →
K. 33 UPRO / 67 UGL (Sharpe-optimal from sweep)	0.808	-63.7%	0.856	-42.0%	$1,615,004
L. 50 SSO / 40 UGL / 10 cash	0.773	-55.7%	0.938	-37.8%	$535,710
M. 50 UPRO / 40 UGL / 10 cash	0.716	-71.7%	0.967	-50.5%	$912,190
N. 40 UPRO / 50 UGL / 10 cash	0.775	-64.4%	0.936	-44.7%	$1,037,421
O. 35 UPRO / 45 UGL / 10 BTC / 10 cash	1.210 (11.6y)	-42.9%	—	—	$262,139
P. 30 SSO / 50 UGL / 10 BTC / 10 cash	1.216 (11.6y)	-32.6%	—	—	$168,793
Q. 50 UPRO / 30 UGL / 10 TQQQ / 10 cash	0.645	-79.0%	0.986	-55.3%	$759,317

Sensitivity: UPRO/UGL ratio sweep (stress window 2000-2026)

UPRO %	UGL %	CAGR	MDD	Sharpe	Calmar	$10k →
0%	100%	19.56%	-74.4%	0.681	0.263	$969,784
25%	75%	21.76%	-60.1%	0.805	0.362	$1,562,502
33%	67%	21.92%	-63.7%	0.808	0.344	$1,616,542
50%	50%	21.35%	-71.6%	0.748	0.298	$1,433,605
60%	40%	20.44%	-79.2%	0.690	0.258	$1,180,823
67%	33%	19.54%	-83.5%	0.649	0.234	$975,715
75%	25%	18.27%	-87.5%	0.604	0.209	$741,763
80%	20%	17.35%	-89.6%	0.579	0.194	$606,074
90%	10%	15.19%	-92.9%	0.534	0.164	$376,599
100%	0%	12.65%	-96.4%	0.497	0.131	$212,319

Sharpe peaks at ~25-33% UPRO / 67-75% UGL. Heavy gold weighting consistently beats heavy equity on risk-adjusted basis across the 25-year window.

BTC sleeve sensitivity (BTC-era 2014-09 → 2026-05)

Adding BTC progressively into a 60% UPRO + 40% UGL base, sliding the equity sleeve to make room:

UPRO %	UGL %	BTC %	CAGR	MDD	Sharpe	$10k →
60%	40%	0%	32.43%	-57.0%	0.992	$262,825
55%	40%	5%	34.77%	-54.9%	1.075	$322,431
50%	40%	10%	36.98%	-52.7%	1.153	$389,513
45%	40%	15%	39.04%	-50.5%	1.220	$463,437
40%	40%	20%	40.95%	-48.3%	1.271	$543,138
35%	40%	25%	42.70%	-48.4%	1.304	$627,099

BTC monotonically improves Sharpe and reduces MDD over its 11.6-year window. Each 5% BTC slice adds ~0.05-0.08 Sharpe.

Interpretation

Most counterintuitive finding: heavy gold beats heavy equity on long-window Sharpe

The Sharpe-optimal UPRO/UGL ratio over 25 years is ~25-33% UPRO / 67-75% UGL. Heavy GOLD weighting, not heavy equity. the heavy-equity hypothesis of 75/25 UPRO/UGL gives Sharpe 0.604 — meaningfully WORSE than 25/75 UPRO/UGL's 0.805.

Why: Over 25 years including dotcom + GFC + 2022, gold (Sharpe 0.71 at 1x, 0.68 at 2x) had higher risk-adjusted returns than equity (SPY 1x Sharpe 0.51, 3x SPY synthetic 0.50). Combining two assets, weighting more toward the higher-Sharpe asset improves the combination. Gold defended in exactly the regimes when leveraged equity wiped out.

The CAGR ranking is different: 33% UPRO / 67% UGL hits 21.92% CAGR vs 75/25's 18.27%. So heavy gold also wins on absolute CAGR over the stress window, not just risk-adjusted.

This is partly regime-specific. Gold had two massive bull runs in this window (2001-2011, 2019-present). A window without those would show less gold dominance. But the same dynamic shows up over 86 years of S&P data in our earlier studies — gold has a long history of being structurally good defensive when equities crash.

The heavy-leveraged-with-metals portfolio holds up surprisingly well

Current (A) gets Sharpe 0.840 on the stress window in this study. Note the caveat: this used TLT as ZROZ proxy, so the window starts 2002 (skipping dotcom's worst months). On apples-to-apples 25.7-year synthetic comparison (the earlier portfolio_backtest study), current portfolio Sharpe was 0.589.

The diversification across 9 assets does meaningful work. Even with the questioned holdings (USD, SOXL, AGQ, ZROZ), the combination is reasonable.

Could the same Sharpe be achieved with a simpler portfolio? Yes — portfolio K (33% UPRO + 67% UGL) gets 0.808 with much simpler structure. But that's a higher-conviction bet on gold continuing to outperform, less diversification, more concentrated risk if gold reverses.

2x (SSO) versions consistently have the best MDD

Portfolio I (60 SSO + 30 UGL + 10 cash): Sharpe 0.709, -60.73% MDD over 25 years. That's the lowest MDD of any high-Sharpe portfolio tested.

For someone optimizing for "stay invested through any regime," the 2x version beats the 3x version on the most relevant trade-off (drawdown-adjusted return). Calmar 0.260 for portfolio I beats most of the 3x portfolios except the heavy-gold ones.

The implication: if you can't psychologically survive -80% drawdowns, the 2x version is structurally better. The CAGR difference (15.8% vs 18-22%) is real but smaller than the MDD savings would suggest.

BTC has real evidence — but the window is structurally questionable

Adding BTC to a leveraged portfolio improves every metric we measure over its 11.6-year history. Sharpe 0.992 (no BTC) → 1.304 (25% BTC). MDD -57% → -48%. CAGR 32% → 43%.

The caveat is enormous: BTC's entire 11.6-year history is QE/ZIRP era + AI bull. We don't have a single example of BTC behavior during a sustained high-rate regime, a deflationary recession, or a prolonged equity-and-commodities-down period. The Sharpe of 1.0+ is partly a structural artifact of the regime BTC has lived through, not a forward-looking expected value.

Honest framing: BTC at 5-15% allocation provides a real-and-meaningful Sharpe lift in backtests over the available window. Whether that lift persists going forward is a regime-bet, not a backtest-validated bet. The same logic that says "the trend filter doesn't predict the next decade from past decades" applies to BTC's Sharpe.

Silver fails on every metric

Portfolio E (with 15% SILV) gets stress Sharpe 0.676 vs portfolio C's 0.647 (without silver). Marginal. And silver actively hurt every other portfolio it was added to. Combined with the durable finding that silver leveraged is a wipeout across windows: silver at any allocation level adds minimal value over cash.

Portfolio H (no defensive, just leveraged equities + gold) is the bull-window winner

Portfolio H (50 UPRO + 25 UGL + 25 TQQQ) hits Sharpe 0.990 and $942k on the bull window. Best dollar return of any portfolio tested. But on the stress window: Sharpe drops to 0.591 with -90.4% MDD. Massive regime exposure.

If you're confident in continued bull regime, H is the dollar-maximizer. If you want defense across regimes, it's the worst risk-adjusted choice in this study.

Practical synthesis (data only, you decide)

Here's what the data suggests, ranked by what each metric prioritizes:

If you optimize for Sharpe across multiple regimes: 25-33% leveraged equity (UPRO or SSO) + 67-75% UGL. Heavy gold weighting. Possibly 5-15% BTC sleeve if you accept regime risk. No silver. No ZROZ.

If you optimize for drawdown-adjusted return (Calmar): Portfolio K (33 UPRO / 67 UGL) or P (30 SSO / 50 UGL / 10 BTC / 10 cash with BTC era only). Both heavy gold, conservative equity sleeve.

If you optimize for absolute dollars in a continued bull: Portfolio H (50 UPRO / 25 UGL / 25 TQQQ). Highest historical end value on bull window but worst stress-window protection.

If you want to keep something close to your current portfolio but address the documented weaknesses: drop SOXL (3x semis = structurally bad), reduce or eliminate AGQ (silver leveraged = structurally bad), reduce or eliminate ZROZ (TLT defensive failed in 2022), reallocate toward UGL. Even a partial shift in this direction would improve historical risk-adjusted metrics.

The portfolio you're currently running scores well on the bull window (Sharpe ~0.96) but is in the middle of the pack on the stress window. Whether to optimize for the regime ahead or for robustness across regimes is your call — the data can show you the trade-off but not which side of it to land on.

Caveats

Synthetic LETFs ignore real borrow cost. Pre-2009 LETFs paid 4-15% borrow rates. The synthetic numbers understate real costs.
BTC's 11.6-year history is regime-incomplete. Sharpe 1.0+ in BTC-inclusive portfolios is partly a regime artifact.
Gold's 25-year history includes two major bull runs. A window without 2001-2011 and 2019-present gold bulls would show less gold dominance.
No tax modeling. Weekly rebalance generates short-term capital gains in taxable accounts. After-tax results would push toward less-frequent rebalancing or lower-turnover portfolios.
Current portfolio (A) test starts 2002, not 2000. TLT inception limits the start date. The 0.840 Sharpe is NOT directly comparable to the 0.589 from the earlier stress test that used a duration-based synthetic for ZROZ pre-2002.
All allocations are static — no trend filtering applied. The earlier studies showed SMA200 filtering equity (not metals) adds Sharpe. None of these portfolios use that. Adding SMA200 filtering to the UPRO sleeve could meaningfully improve some of these results — separate study.

Source

Saved log: /tmp/portfolio_search.log.

Inline runner: see a sample heavy-leveraged portfolio (internal-only study) for the same pattern; this study adds the candidate-portfolio loop and the BTC sleeve.

a sample heavy-leveraged portfolio (internal-only study) — baseline performance of current portfolio
Full leverage spectrum — per-underlying analysis informing why heavy gold wins
Defensive bucket comparison — why ZROZ/TLT failed as defensive
LRS family replication + walk-forward — caution on SMA-trend-filter overfitting

Open questions for next research

~~Apply SMA200 to the leveraged equity sleeve of each portfolio~~ DONE — see follow-up below.
Walk-forward test on the UPRO/UGL ratio — is the 25-33% UPRO sweet spot robust if you tune on 2000-2015 and evaluate on 2016-2026? Or is it a fit to the specific gold bull windows?
Tax-after analysis — how do the rankings shift with realistic short-term gains tax on weekly rebalance?
2x-only portfolio walkforward — Portfolio I's combination of low MDD + decent Sharpe is interesting. A deeper study on 2x-only combinations might find more sweet spots.

FOLLOW-UP (same session): SMA200 filter applied to equity sleeve only

Question: if we apply SMA200 filtering to the leveraged-equity sleeve only (UPRO, SSO, TQQQ) and leave gold/BTC/cash unfiltered, does it improve any of the portfolio archetypes?

Methodology: When SPY > SMA200, hold each asset at target weight. When SPY < SMA200, equity assets (UPRO/SSO/TQQQ) get routed to CASH; gold/BTC/cash sleeves unchanged. Signal lagged 1 bar. Same weekly rebalance, same 1bp/side costs.

Results: STRESS window (25.7 years, 2000-2026)

Portfolio	B&H Sharpe	Filter Sharpe	Δ Sharpe	B&H MDD	Filter MDD	Δ MDD	Filter $10k →
B. 75 UPRO / 25 UGL	0.604	0.861	+0.257	-87.5%	-37.7%	+49.8pp	$1,619,056
K. 33 UPRO / 67 UGL	0.808	0.890	+0.082	-63.7%	-40.3%	+23.5pp	$1,752,746
N. 40 UPRO / 50 UGL / 10 cash	0.775	0.934	+0.160	-64.4%	-31.9%	+32.6pp	$1,195,923
L. 50 SSO / 40 UGL / 10 cash	0.773	0.933	+0.159	-55.7%	-26.1%	+29.6pp	$538,721
H. 50 UPRO / 25 UGL / 25 TQQQ	0.591	0.877	+0.286	-90.4%	-37.4%	+53.1pp	$2,354,978
O. 35 UPRO / 45 UGL / 10 BTC / 10 cash	1.210	1.254	+0.044	-42.9%	-27.1%	+15.8pp	$173,845

Results: BULL window (14.3 years, 2012-2026)

The filter slightly hurts on the bull window (-0.02 to -0.05 Sharpe, ~-3 to -8pp CAGR) because the filter sometimes routes equity to cash on minor corrections that then immediately recover. That's the price of insurance. MDDs still improve.

Interpretation:

Every portfolio improves on Sharpe with the filter applied to equity. The previous "heavy gold beats heavy equity" finding gets neutralized — with the filter, heavy-equity portfolios (B, H) now match or beat heavy-gold portfolios on Sharpe AND on drawdown.

The clear winners on the stress window:

N + filter (40 UPRO / 50 UGL / 10 cash + SMA on equity): Sharpe 0.934 — HIGHEST non-BTC Sharpe in the entire study
L + filter (50 SSO / 40 UGL / 10 cash + SMA on equity): Sharpe 0.933, MDD just -26.1% — best MDD-adjusted return
H + filter (50 UPRO / 25 UGL / 25 TQQQ + SMA on both LETFs): $2.35M from $10k over 25 years — absolute dollar winner

The "best of both worlds" pattern: heavy gold for defense + return, SMA200 on equity for catastrophic-drawdown protection. Portfolios that combine these two ideas dominate every other configuration tested.

Caveats specific to this follow-up:

Single SMA200 signal applied to all equity assets — SPY's SMA200 routes UPRO, SSO, AND TQQQ to cash. A QQQ-based signal for TQQQ might perform marginally differently.
No tolerance band or RSI overlay — pure mechanical SMA200 cross. Earlier walk-forward studies showed parameter tuning doesn't transfer; pure SMA stays robust.
Bull window cost is real — filter underperforms B&H by 3-8pp CAGR in the 2012-2026 stretch. Real cost during clean uptrends.

Saved log: /tmp/portfolio_with_filter.log.

Consolidated comparison: ALL 17 portfolios, B&H vs SMA-on-equity, both windows

Stress window 2000-2026 (or asset-available start for BTC portfolios)

Portfolio	Yrs	B&H CAGR	B&H MDD	B&H Sharpe	SMA CAGR	SMA MDD	SMA Sharpe	Δ Sharpe
A. Current (9-asset)	23.8	25.32%	-69.54%	0.840	22.11%	-34.65%	0.913	+0.073
B. 75 UPRO / 25 UGL	25.7	18.27%	-87.48%	0.604	21.93%	-37.70%	0.861	+0.257
C. 50 UPRO / 25 UGL / 25 cash	25.7	15.86%	-72.55%	0.647	17.02%	-26.42%	0.905	+0.258
D. 60 UPRO / 20 UGL / 20 AGQ	25.7	19.96%	-81.71%	0.654	22.64%	-43.98%	0.843	+0.189
E. 50 UPRO / 25 UGL / 15 SILV / 10 cash	25.7	17.76%	-73.16%	0.676	19.14%	-30.48%	0.901	+0.226
F. 40 UPRO / 20 TQQQ / 20 UGL / 10 AGQ / 10 cash	25.7	18.32%	-83.75%	0.624	21.95%	-35.07%	0.887	+0.263
G. 40 UPRO / 20 UGL / 20 BTC / 20 cash	11.6	35.31%	-46.16%	1.217	30.17%	-30.79%	1.294	+0.077
H. 50 UPRO / 25 UGL / 25 TQQQ	25.7	18.12%	-90.41%	0.591	23.72%	-37.35%	0.877	+0.286
I. 60 SSO / 30 UGL / 10 cash	25.7	15.79%	-60.73%	0.709	16.03%	-24.08%	0.932	+0.224
J. 40 UPRO / 20 UGL / 20 cash / 20 TQQQ	25.7	16.06%	-83.26%	0.587	19.31%	-30.79%	0.875	+0.288
K. 33 UPRO / 67 UGL	25.7	21.92%	-63.73%	0.808	22.31%	-40.28%	0.890	+0.082
L. 50 SSO / 40 UGL / 10 cash	25.7	16.79%	-55.75%	0.774	16.81%	-26.11%	0.933	+0.159
M. 50 UPRO / 40 UGL / 10 cash	25.7	19.23%	-71.74%	0.716	20.50%	-29.35%	0.941	+0.225
N. 40 UPRO / 50 UGL / 10 cash	25.7	19.83%	-64.45%	0.775	20.50%	-31.88%	0.935	+0.160
O. 35 UPRO / 45 UGL / 10 BTC / 10 cash	11.6	32.39%	-42.86%	1.210	27.80%	-27.08%	1.255	+0.044
P. 30 SSO / 50 UGL / 10 BTC / 10 cash	11.6	27.48%	-32.63%	1.216	24.65%	-26.24%	1.211	-0.005
Q. 50 UPRO / 30 UGL / 10 TQQQ / 10 cash	25.7	18.38%	-79.04%	0.645	21.03%	-31.05%	0.917	+0.272

Bull window 2012-2026 (14.3 years, or 11.6 for BTC portfolios)

Portfolio	B&H Sharpe	SMA Sharpe	Δ Sharpe
A. Current (9-asset)	0.961	0.911	-0.049
B. 75 UPRO / 25 UGL	0.952	0.977	+0.025
C. 50 UPRO / 25 UGL / 25 cash	0.965	0.986	+0.022
D. 60 UPRO / 20 UGL / 20 AGQ	0.882	0.873	-0.008
E. 50 UPRO / 25 UGL / 15 SILV / 10 cash	0.939	0.947	+0.008
F. 40 UPRO / 20 TQQQ / 20 UGL / 10 AGQ / 10 cash	0.964	0.971	+0.008
G. 40 UPRO / 20 UGL / 20 BTC / 20 cash	1.217	1.294	+0.077
H. 50 UPRO / 25 UGL / 25 TQQQ	0.990	1.009	+0.019
I. 60 SSO / 30 UGL / 10 cash	0.964	0.972	+0.008
J. 40 UPRO / 20 UGL / 20 cash / 20 TQQQ	0.989	1.007	+0.018
K. 33 UPRO / 67 UGL	0.857	0.811	-0.045
L. 50 SSO / 40 UGL / 10 cash	0.938	0.917	-0.021
M. 50 UPRO / 40 UGL / 10 cash	0.967	0.974	+0.007
N. 40 UPRO / 50 UGL / 10 cash	0.936	0.915	-0.021
O. 35 UPRO / 45 UGL / 10 BTC / 10 cash	1.210	1.255	+0.044
P. 30 SSO / 50 UGL / 10 BTC / 10 cash	1.216	1.211	-0.005
Q. 50 UPRO / 30 UGL / 10 TQQQ / 10 cash	0.986	1.008	+0.021

Key observations from the consolidated comparison:

Filter helps EVERY portfolio on stress window (Δ Sharpe +0.04 to +0.288, MDD reduction 10-50pp). Pattern is universal — gold-heavy portfolios get smaller Sharpe boost but already had high baselines; equity-heavy portfolios get the biggest boost.
Top 5 non-BTC stress-window Sharpe, all with SMA on equity: - M (50 UPRO / 40 UGL / 10 cash): 0.941 - N (40 UPRO / 50 UGL / 10 cash): 0.935 - L (50 SSO / 40 UGL / 10 cash): 0.933 - I (60 SSO / 30 UGL / 10 cash): 0.932 - Q (50 UPRO / 30 UGL / 10 TQQQ / 10 cash): 0.917
Highest stress-window CAGR (with filter): H (50 UPRO / 25 UGL / 25 TQQQ) at 23.72%, $2.35M from $10k. Lower Sharpe (0.877) and worse MDD (-37%) than M-class portfolios; aggressive choice.
Lowest stress-window MDD (with filter): I (60 SSO / 30 UGL / 10 cash) at -24.08%. Best drawdown protection of any portfolio tested. CAGR is lower (16%) but the risk-adjusted profile is excellent.
Current portfolio (A) + filter gets Sharpe 0.913 with MDD -34.65%. Solid improvement on its own B&H (Sharpe 0.840, MDD -69.5%). The 9-asset diversification still works fine; adding the SMA200 filter to the equity sleeve is by far the most impactful single change.
Bull window: filter slightly helps or is neutral for most portfolios. Net cost is modest. The trade-off (small bull-window cost, large stress-window gain) is universally favorable across the candidate set.