WF05The viral 50, re-tested · Daily
Not re-testedOptuna TPE Optimization + Walk-Forward
The rules
Searches SMA windows with Optuna; best was SMA(23/206).
- Universe
- XLB, XLE, XLF, XLI, XLK, XLP, XLU, XLV and XLY
- Bars
- daily
- Position size
- -
- Rebalance
- -
Results
| Measure | Orbital re-test | Published | SPY, same days |
|---|---|---|---|
| Window | Not re-tested | — | — |
| Total return | — | — | — |
| CAGR | — | — | — |
| Sharpe | — | — | — |
| Discounted Sharpe | — | — | — |
| Max drawdown | — | — | — |
| Win rate | — | — | — |
| Profit factor | — | — | — |
| Trades | — | — | — |
Rebuild it in Orbital
Paste this into Orbital Chat. The Strategy Developer builds the workflow and asks before it runs any backtest.
Closest Orbital version: Orbital has no built-in parameter search, so this tests 23/206, the pair the original's 30 Optuna trials picked on its equal-weight version from 2000 to 2025, against the five pairs around it, none slower than 206 days because a slower SMA could not warm up on data that begins in January 2016. Orbital's discount counts only these six pairs, not the 30 trials, and those trials ran on data covering most of this window, so even the discounted Sharpe flatters 23/206; for a walk-forward check of the winner like the original's, run WF01's prompt with 23 and 206 days in place of 50 and 200.
Trade the sector ETFs XLB, XLE, XLF, XLI, XLK, XLP, XLU, XLV and XLY on daily bars. Buy an ETF with 11% of equity when its fast SMA is above its slow SMA, and sell it when the fast SMA falls below the slow SMA. Long only; uninvested capital stays in cash. Build it as one strategy with the two SMA lengths as settings, then test the pair an earlier parameter search chose, a 23-day fast SMA with a 206-day slow SMA, against the pairs around it: backtest every pair of an 18-, 23- or 28-day fast SMA with a 186- or 206-day slow SMA from December 15, 2016 to October 30, 2026 with $100,000. That is six backtests. When all six have finished, rank them by Sharpe in one table with annual return and maximum drawdown, say whether 23/206 does clearly better than the pairs around it or about the same, and give its Sharpe after allowing for every pair tried.
Why it was not re-tested
A validation workflow around W01's rule rather than a strategy of its own. Orbital covers part of the same ground for every strategy: its Backtesting list discounts the Sharpe for every configuration tried, and each report resamples the trades 1,000 times.
Sources