This example selects from 23 ETFs covering equities, bonds, REITs and gold. Each week, ETFs above their 120-day exponential moving average are ranked using equally weighted three-month and six-month returns. Up to seven qualifying ETFs form an equally weighted portfolio.
Data-quality notes. Unavailable ranking observations were omitted, factor weights were redistributed over available inputs where necessary, and 15 rebalances with no valid targets were skipped. This uses a fixed ETF universe, not a point-in-time universe.
Performance and strategy calculations use adjusted close, including split and dividend adjustments. Quantities are adjusted simulation units.
Why does it work?
This trend-following strategy aims to participate in equity uptrends and reduce exposure during downtrends by rotating across asset classes.
The strategy changes its asset allocation as market trends change. In the backtest period reported below, it had lower volatility and a smaller maximum drawdown than SPY, but lower annualized returns and a lower Sharpe ratio.
What is the performance?
The Performance and Statistics tabs show the current verified adjusted-price simulation in USD, with SPY as the S&P 500 ETF benchmark. Annualized returns, volatility, Sharpe ratio and maximum drawdown are calculated from the same published result. Use the chart dates to identify the period being compared. The Positions tab samples historical simulated allocations; its last displayed date can differ from the final performance date.
Research paper
Mebane Faber: A Quantitative Approach to Tactical Asset Allocation
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=962461
