Combined PortfolioUSA + Canada + Germany momentum
The three live momentum sleeves — USA (top-7 large caps), Canada (top-10 TSX) and Germany (run at N=10, the top-10 of the published ranking) — combined at 1/3 each, monthly. Common window 2006-03-31 — 2026-04-30 (242 meses) (limited by the USA series start). Each sleeve's own caveats (survivorship windows, capacity, in-sample selection) carry through; see the per-strategy audits.
I. The Case for Combining
| Sleeve | Sharpe | Sortino | CAGR | MaxDD |
|---|---|---|---|---|
| US | 0.92 | 1.94 | +25.6% | -30.3% |
| CA | 1.27 | 3.32 | +34.3% | -24.5% |
| DE | 0.76 | 1.45 | +18.4% | -39.1% |
| COMBINED 1/3 each | 1.52 | 3.67 | +27.5% | -16.6% |
The combined portfolio's Sharpe (1.52) exceeds every individual sleeve's, and its MaxDD (−16.6%) is barely half the best individual sleeve's. That is the diversification premium of three strategies with 0.19–0.25 correlations and regimes that de-risk independently.
II. Correlations
| DE | CA | US | |
|---|---|---|---|
| DE | 1.00 | 0.19 | 0.244 |
| CA | 0.19 | 1.00 | 0.247 |
| US | 0.244 | 0.247 | 1.00 |
Full-series monthly correlations. Two structural sources: different stock selection (measured 0.23–0.45 when both sleeves are invested) plus regime divergence — roughly a quarter of all months have one sleeve in cash while another is invested, and the invested sleeve earns above its average in those months. Caveat: rolling 36-month correlations are unstable — DE-CA has ranged -0.14 to 0.57, DE-US -0.13 to 0.62, CA-US -0.21 to 0.64. Do not budget on the averages holding in every triennium.
Pairwise combos (what each sleeve adds)
| Pair (50/50) | Sharpe | Sortino | CAGR | MaxDD |
|---|---|---|---|---|
| DE+CA | 1.4 | 3.39 | +27.3% | -25.4% |
| DE+US | 1.13 | 2.34 | +23.0% | -21.7% |
| CA+US | 1.45 | 3.52 | +31.2% | -16.3% |
III. Monthly Return Distribution
Mean +2.15%/month, standard deviation 4.66% (annualized ~25.8% / 16.1% vol). 62% of months positive, skew +0.78 — the fat tail is on the upside. Worst month in 242 meses: -7.52%; best: +20.75%.
- Normal month (±1σ): -2.5% to +6.8% — about 2 of every 3 months.
- Bad month (−2σ): ~-7.2% — roughly once every 2 years.
- Outside the entire 20-year sample: worse than −10% in a month — never happened; treat as an investigation trigger, not noise.
IV. Worst Months — How It Fails
| Month | Combo | DE | CA | US |
|---|---|---|---|---|
| 2018-10-31 | -7.5% | -0.0% | -0.0% | -22.5% |
| 2026-03-31 | -7.1% | 3.5% | -15.6% | -9.1% |
| 2020-02-29 | -6.7% | -3.8% | -10.7% | -5.7% |
| 2006-05-31 | -6.5% | -7.9% | -7.4% | -4.3% |
| 2019-09-30 | -6.5% | 2.0% | -9.0% | -12.5% |
| 2007-11-30 | -6.5% | -10.5% | -4.9% | -4.0% |
| 2014-07-31 | -6.1% | -13.4% | -2.6% | -2.2% |
| 2008-01-31 | -5.7% | -16.9% | -0.2% | 0.0% |
| 2015-08-31 | -5.5% | -9.4% | -0.2% | -7.0% |
| 2011-08-31 | -5.3% | -8.1% | -0.1% | -7.7% |
The failure mode is always one sleeve blowing up while the others cushion (US −22.5% in Oct-2018 → combo −7.5%; CA −15.6% in Mar-2026 → combo −7.1%). In 242 months the three sleeves never all fell hard together — the common regime mechanism pulls them all to cash in genuine crises instead.
Best Months — How It Wins
| Month | Combo | DE | CA | US |
|---|---|---|---|---|
| 2021-02-28 | +20.7% | +23.8% | +39.9% | -1.4% |
| 2024-02-29 | +18.8% | +11.8% | +8.5% | +36.1% |
| 2025-09-30 | +16.3% | +11.0% | +14.5% | +23.3% |
| 2017-01-31 | +15.1% | +11.7% | +27.1% | +6.5% |
| 2026-04-30 | +14.5% | -0.2% | +5.9% | +37.8% |
| 2018-08-31 | +13.1% | -1.4% | +23.4% | +17.4% |
| 2024-01-31 | +12.7% | +4.1% | +18.0% | +16.1% |
| 2020-12-31 | +12.5% | +13.4% | +15.0% | +9.1% |
| 2026-01-31 | +11.8% | +3.0% | +10.1% | +22.1% |
| 2020-06-30 | +10.7% | -0.1% | +21.4% | +10.7% |
V. Drawdown Profile
MaxDD -16.6%. Time spent in >5% drawdown: 28% of months (normal operation, not a malfunction); in >10%: 9%. All drawdown episodes deeper than 8%:
| Start | Trough | Depth |
|---|---|---|
| 2006-06-30 | 2006-09-30 | -13.0% |
| 2008-01-31 | 2009-04-30 | -16.6% |
| 2010-06-30 | 2010-06-30 | -9.4% |
| 2011-08-31 | 2011-12-31 | -9.5% |
| 2018-12-31 | 2019-01-31 | -10.5% |
| 2021-09-30 | 2021-09-30 | -8.3% |
| 2023-05-31 | 2023-10-31 | -13.1% |
VI. Weight Sensitivity
| Weights | Sharpe | Sortino | CAGR | MaxDD |
|---|---|---|---|---|
| 1/3 cada (baseline) | 1.52 | 3.67 | +27.5% | -16.6% |
| DE25/CA40/US35 | 1.54 | 3.82 | +28.7% | -16.0% |
| DE20/CA50/US30 | 1.55 | 3.94 | +29.9% | -17.1% |
| DE50/CA25/US25 | 1.42 | 3.25 | +25.5% | -19.3% |
| DE25/CA25/US50 | 1.42 | 3.34 | +27.3% | -15.4% |
| sem DE (50/50 CA+US) | 1.45 | 3.52 | +31.2% | -16.3% |
Equal weight is within 0.03 Sharpe of every sensible tilt — the diversification does the work, not the weights. Dropping Germany entirely costs 0.07 Sharpe and ~1.6pp of MaxDD; overweighting any single sleeve to 50% costs ~0.10. There is nothing to optimize here.
VII. Calendar Years
| Year | Combo | US | CA | DE |
|---|---|---|---|---|
| 2006 | +17.3% | -3.7% | +41.1% | +13.9% |
| 2007 | +36.8% | +79.4% | +40.6% | -1.3% |
| 2008 | -10.2% | +0.0% | -13.0% | -18.2% |
| 2009 | +25.3% | +9.8% | +30.2% | +34.3% |
| 2010 | +34.6% | +12.3% | +54.2% | +37.0% |
| 2011 | -5.4% | -3.4% | +3.6% | -16.8% |
| 2012 | +29.9% | +21.1% | +30.7% | +36.8% |
| 2013 | +73.1% | +64.1% | +84.4% | +68.0% |
| 2014 | +16.5% | +27.9% | +17.6% | +2.3% |
| 2015 | +26.9% | +23.3% | +11.5% | +46.9% |
| 2016 | +30.4% | +18.4% | +78.2% | +2.1% |
| 2017 | +64.3% | +37.2% | +79.2% | +69.7% |
| 2018 | +9.0% | -10.0% | +25.2% | +8.2% |
| 2019 | +21.7% | +3.7% | +16.1% | +47.1% |
| 2020 | +47.5% | +71.8% | +50.6% | +18.7% |
| 2021 | +23.3% | +20.4% | +29.5% | +12.6% |
| 2022 | +6.9% | +0.2% | +25.8% | -7.9% |
| 2023 | -5.6% | +4.0% | -5.9% | -16.2% |
| 2024 | +72.5% | +116.2% | +63.0% | +37.8% |
| 2025 | +49.6% | +14.8% | +82.7% | +48.4% |
| 2026 | +31.3% | +88.7% | +5.4% | +7.3% |
Mean year +28.4%; 3 negative years in 21 (worst: -10.2% in 2008); best +73.1% (2013).
VIII. Rolling 3-Year Performance
Mean rolling 3-year CAGR +25.9%, range +7.6% to +50.2%, negative in 0.0% of all 3-year windows.
IX. Honest Forward Expectations
Backtest numbers are in-sample compositions of in-sample strategies. Applying the standard live haircut (expected live Sharpe ≈ 0.6–0.75× backtest): expect roughly 15–20%/year at Sharpe ~1.0–1.2, and budget for a real MaxDD around −20 to −25% rather than the backtest's −16.6%. Judge the portfolio against a month-by-month simulated track, over 3+ year horizons — any single year between −10% and +40% is within script.
XI. Germany Sleeve: N=10 vs N=20
| Variant | Sharpe | CAGR | Ann. vol | MaxDD |
|---|---|---|---|---|
| Germany sleeve N=10 (standalone) | 0.76 | +18.4% | 20.2% | -39.1% |
| Germany sleeve N=20 (standalone) | 0.88 | +17.4% | 16.3% | -27.8% |
| Combo with DE N=10 (operational choice) | 1.52 | +27.5% | 16.1% | -16.6% |
| Combo with DE N=20 | 1.54 | +27.0% | 15.5% | -15.2% |
Standalone, N=10 is clearly worse for the German sleeve (MaxDD −39% vs −28%). At the combo level the difference nearly vanishes — 0.02 of Sharpe and 1.4pp of MaxDD — because the extra idiosyncratic risk diversifies away (N=10's correlations to the other sleeves are even slightly lower). N=10 is the operational choice: buy the top-10 of the published top-20 at 10% each.
Distribution detail (combo)
| DE N=10 | DE N=20 | |
|---|---|---|
| Mean monthly return | +2.15% | +2.11% |
| Monthly standard deviation | 4.66% | 4.48% |
| Annualized volatility | 16.1% | 15.5% |
| Worst month | -7.5% | -8.2% |
| Best month | +20.7% | +18.4% |
| Months positive | 62% | 62% |
XII. Monthly Returns Over Time
Every monthly return of the combo (DE N=10 variant), with ±1σ (dashed) and ±2σ (dotted) bands around the mean. Full month-by-month table:
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2006 | +7.0 | +3.8 | -6.5 | -3.7 | -0.2 | +1.6 | -4.7 | +8.6 | +1.5 | +10.2 | +17.3 | ||
| 2007 | +2.6 | +0.3 | +2.3 | +6.1 | +7.7 | +2.7 | +1.9 | +0.6 | +3.5 | +9.1 | -6.5 | +2.4 | +36.8 |
| 2008 | -5.7 | -0.1 | -0.5 | +0.2 | +0.9 | -1.2 | -3.8 | -0.1 | +-0.0 | -0.1 | -0.1 | -0.1 | -10.2 |
| 2009 | -0.1 | -0.1 | -0.1 | -2.8 | +5.6 | +0.9 | +3.6 | +1.8 | +6.8 | +0.4 | +7.4 | -0.2 | +25.3 |
| 2010 | -1.9 | +5.6 | +9.7 | +4.8 | -5.2 | -4.4 | +4.2 | +1.1 | +2.8 | +6.1 | +5.7 | +2.5 | +34.6 |
| 2011 | -1.4 | +5.9 | -0.9 | +0.2 | -0.9 | -1.8 | +-0.0 | -5.3 | -0.1 | -0.1 | +-0.0 | -0.8 | -5.4 |
| 2012 | +3.2 | +6.6 | +3.0 | +1.2 | -3.5 | +0.4 | +3.0 | +1.7 | +3.0 | +1.2 | +4.5 | +2.5 | +29.9 |
| 2013 | +7.2 | +4.3 | +6.5 | +0.5 | +2.6 | -2.0 | +8.4 | +2.9 | +9.6 | +10.1 | +4.4 | +2.4 | +73.1 |
| 2014 | +2.2 | +6.0 | +3.5 | -1.6 | +6.5 | -0.2 | -6.1 | +2.0 | +1.6 | -1.1 | +2.3 | +1.0 | +16.5 |
| 2015 | +4.7 | +9.1 | +6.4 | -1.7 | +6.7 | -0.3 | +3.4 | -5.5 | -0.1 | -0.1 | +1.1 | +1.1 | +26.9 |
| 2016 | -0.1 | +-0.0 | +1.2 | +4.0 | +1.9 | +7.7 | +5.6 | -1.0 | +4.0 | -3.8 | +3.1 | +4.8 | +30.4 |
| 2017 | +15.1 | +1.2 | +3.4 | +1.4 | +9.0 | -0.6 | +0.8 | -1.0 | +6.3 | +4.0 | +3.6 | +8.7 | +64.3 |
| 2018 | +0.9 | +3.6 | -2.0 | +0.3 | +4.2 | +5.1 | -4.9 | +13.1 | +0.6 | -7.5 | -0.1 | -3.1 | +9.0 |
| 2019 | -0.1 | +1.7 | +4.8 | -0.9 | +1.8 | +3.0 | +0.9 | +1.4 | -6.5 | +4.1 | +3.3 | +6.9 | +21.7 |
| 2020 | +0.8 | -6.7 | -0.1 | -0.1 | -0.1 | +10.7 | +7.2 | +9.4 | +5.6 | -2.3 | +4.6 | +12.5 | +47.5 |
| 2021 | +10.6 | +20.7 | +0.1 | -2.9 | +1.5 | -0.9 | -3.7 | -0.8 | -1.6 | +5.3 | -3.8 | -0.8 | +23.3 |
| 2022 | +2.5 | +5.1 | -0.1 | +1.3 | +0.0 | +-0.0 | -0.1 | -0.1 | -0.1 | -0.1 | +0.9 | -2.5 | +6.9 |
| 2023 | +0.1 | -5.2 | +1.7 | -2.2 | -4.0 | +4.0 | +5.0 | -5.1 | -1.7 | -3.3 | +3.6 | +2.2 | -5.6 |
| 2024 | +12.7 | +18.8 | +7.2 | +1.2 | +9.3 | -1.5 | -3.5 | +1.5 | +6.7 | +2.9 | +7.5 | -4.7 | +72.5 |
| 2025 | +7.8 | -1.3 | +0.7 | +3.4 | +2.8 | -1.0 | +9.9 | +2.8 | +16.3 | +1.1 | -4.5 | +4.6 | +49.6 |
| 2026 | +11.8 | +10.4 | -7.1 | +14.5 | +31.3 |
X. Limitations
- Common window starts 2006 (USA series); Canada/Germany sleeve numbers over this window differ slightly from their full-sample (2000+) headline figures.
- Each sleeve's own caveats compound here: Canada/Germany pre-2015-ish survivorship, capacity limits (CA/DE sub-$2M sleeves), multi-market selection bias.
- Correlations are regime-dependent — the 0.2 averages have ranged to 0.65 in 3-year stretches.
- Germany sleeve here is N=10 (top-10 of the published top-20, 10% each) — the operational choice; N=20 gives a near-identical combo (Sharpe 1.54, MaxDD −15.2%).