Graveyard / Momentum / ETFs
Country and sector momentum rotation
Does buying last year's winning countries or sectors, and shorting the losers, work?
The idea
Countries and industries that did best over the past 6 to 12 months tend to keep outperforming for a while. Rank them monthly, buy the top and short the bottom.
How we tested it
| Data | Test 1: 20 iShares country ETFs, rank by 12-1 month return, long the top quartile and short the bottom, equal weight and dollar-neutral, monthly, plus a long-only variant. Test 2: the 9 Select Sector SPDRs, long the top 3 and short the bottom 3 on 6-month return, monthly; variants on 1-month formation and on 10 industry ETFs. |
|---|---|
| Period | Test 1: May 1997 to October 2026. Test 2: July 1999 to October 2026; industry ETF variant January 2007 to October 2026. |
| Costs | 5 bp per side, stress 10 bp. No borrow fee in test 1. |
| Holdout | In-sample and out-of-sample Sharpe. |
| Control | An equal-weight basket of all country ETFs, bought and held (Sharpe 0.48). |
Result
| Country long-short: Sharpe full / in-sample / out-of-sample | 0.04 / 0.03 / 0.09 |
|---|---|
| Country long-short: max drawdown | −80% |
| Country long-only top quartile Sharpe | 0.50 |
| Equal-weight country basket Sharpe | 0.48 |
| Sector long-short: Sharpe full / in-sample / out-of-sample | −0.05 / 0.07 / −0.52 |
| Sector long-short: max drawdown | −68% |
Why it failed
Long-short country momentum has had a Sharpe of about zero since 1997 and suffered an −80% momentum crash in 2009. The long-only version is just country beta, matching an equal-weight basket. Industry momentum had a Sharpe of 0.45 before 2000, 0.19 from 2000 and 0.14 from 2013 in the paper check, and was negative in this test.
What would make it worth retesting
No reopen condition is stated for either.
Past results of a backtest say nothing certain about the future.