The rules
At each quarter-end, rank the ETFs by 12-month return and hold the top two, but only those with a positive return.
- Universe
- SPY, EFA, EEM, TLT, IEF, GLD, DBC and VNQ
- Bars
- daily
- Position size
- Equal weight, max 50% each
- On Orbital
- 49% of equity per holding
- Rebalance
- quarter-end
Results
| Measure | Orbital re-test | Published | SPY, same days |
|---|---|---|---|
| Window | Mar 1, 2017 to Oct 30, 2026 | 2007-2025 | Mar 1, 2017 to Oct 30, 2026 |
| Total return | +52.35% | +64.62% | +267.14% |
| CAGR | +4.45% | +2.66% | +14.40% |
| Sharpe | 0.09 | 0.12 | 0.60 |
| Discounted Sharpe | −0.91 | — | — |
| Max drawdown | −42.5% | −42.8% | −33.7% |
| Win rate | 64.9% | 59.5% | — |
| Profit factor | 1.46 | 1.21 | — |
| Trades | 74 | — | — |
Hypothetical: a backtest with modeled costs, not traded with real money.
Hypothetical growth of $10,000
W04 Quarterly Dual Momentum · +52.4% vs SPY +267.1%
Rebuild it in Orbital
Paste this into Orbital Chat. The Strategy Developer builds the workflow and asks before it runs any backtest.
Rotate among SPY, EFA, EEM, TLT, IEF, GLD, DBC and VNQ on daily bars. On the first trading day of each quarter, sell every holding, and the next day buy the two ETFs with the highest 252-day return, 49% of equity each, skipping any whose 252-day return is negative. Long only; uninvested capital stays in cash. Backtest from March 1, 2017 to October 30, 2026 with $100,000.
Is it real?
- Strategies tested together
- 22
- Beta to SPY
- 0.56
- Alpha vs SPY
- −4.6% a year
- Costs paid
- $3,929, 7% of the gain before costs
- Data
- Daily bars, 2,432 sessions
- Fills
- Next open at market, sized at the fill
Read this before you trust it
Orbital never sells a holding because a better one appears. On the first trading day of each quarter the strategy sells everything and buys the new top two at the next day's open, so a name that stays in the top two is sold and bought back. Each holding is 49% of equity, leaving room for costs. Its 12-month lookback needs a year of history, which Orbital's data first has in January 2017, so the first two months of the ten-year window are in cash.
Sources