The strategy does two distinct things at once: it equal-weights the eleven sectors (a tilt away from the cap-weight S&P), and it steps to Treasury bills when the 200-day signal turns down. Which one has been doing the work? To find out, I sliced the record into clean five-year windows from year-end 1999 and separated the two effects exactly.
The full period at a glance
From October 11, 1999 through July 6, 2026 — 26.7 years, one common window, with VBINX as the 60/40 benchmark (its history reaches back far enough to cover the full period):
| Metric | Strategy (EW) | SPY | VBINX (60/40) |
|---|---|---|---|
| Total return | +344.1% | +797.8% | +516% |
| CAGR | 5.74% | 8.56% | 7.1% |
| Max drawdown | −21.5% | −55.2% | −36.0% |
Across every window tested, the story holds: the strategy consistently owns the shallowest drawdown, and the return gap to SPY narrows the further back you go — SPY’s last decade was the outlier.
Returns and drawdowns by five-year slice
Four portfolios in annualized total-return terms. EW B&H is the same eleven equal-weighted sector ETFs, always invested — no timing overlay. The final row is a partial ~1.5-year stub.
| Period | Strategy | VBINX | SPY | EW B&H | Strat DD | VBINX DD | SPY DD |
|---|---|---|---|---|---|---|---|
| YE1999–YE2004 | +1.1% | +2.4% | −2.4% | +2.8% | −22.8% | −25.1% | −47.5% |
| YE2004–YE2009 | +5.6% | +2.9% | +0.4% | +3.2% | −16.6% | −36.0% | −55.2% |
| YE2009–YE2014 | +9.0% | +11.2% | +15.3% | +15.3% | −15.2% | −10.9% | −18.6% |
| YE2014–YE2019 | +4.9% | +7.9% | +11.6% | +9.8% | −10.4% | −11.9% | −19.3% |
| YE2019–YE2024 | +7.9% | +8.1% | +14.4% | +11.9% | −14.2% | −22.8% | −33.7% |
| YE2024–today (stub) | +5.9% | +13.5% | +18.7% | +16.5% | −8.1% | −11.6% | −18.8% |
What drove the return — the attribution
We build three portfolios and add one ingredient at each step. SPY is the neutral base. Going SPY → EW Buy & Hold adds only equal-weighting; going EW B&H → Strategy adds only timing. The subtractions are done in log returns, so the three pieces sum exactly to the strategy’s return with no residual. Units are annualized log-return points per year; the final column converts back to CAGR.
| Period | SPY base | + Equal-wt | + Timing | = Strategy (log) | Strategy CAGR |
|---|---|---|---|---|---|
| YE1999–YE2004 | −2.4% | +5.2% | −1.7% | +1.1% | +1.1% |
| YE2004–YE2009 | +0.4% | +2.7% | +2.3% | +5.4% | +5.6% |
| YE2009–YE2014 | +14.3% | −0.1% | −5.6% | +8.6% | +9.0% |
| YE2014–YE2019 | +10.9% | −1.6% | −4.5% | +4.8% | +4.9% |
| YE2019–YE2024 | +13.5% | −2.2% | −3.6% | +7.6% | +7.9% |
| YE2024–today (stub) | +17.1% | −1.8% | −9.6% | +5.7% | +5.9% |
Reading it
The strategy only out-returns the field in the one slice that contains a major crash — YE2004–YE2009 (+5.6%/yr vs SPY +0.4%, VBINX +2.9%), where timing added +2.3 points a year by sidestepping 2008. In every bull-market slice the timing overlay is a drag of roughly 4–10 points a year — the running cost of the insurance. That cost buys the much shallower drawdowns: in the two crisis slices the strategy fell −23% and −17% while SPY fell −48% and −55%. Equal-weighting was a strong tailwind in the post-bubble early 2000s (+5.2 pts/yr) but faded to a slight drag once mega-cap tech began leading the cap-weight index in the 2010s.
Methodology, briefly
Signal: 200-day SMA with a ±1% hysteresis band on split-adjusted closes. Returns: total-return (dividend-reinvested) series; out-of-market cash earns the actual 3-month T-bill rate, daily-compounded; equal weight across the eleven sector SPDRs, growing as they launched (nine from 1999, XLRE from July 2016, XLC from April 2019). Drawdowns from daily data. No transaction costs or taxes modeled; the final stub slice is only ~1.5 years, so its figures are noisier than the full five-year slices.
— John