Quantitative Strategies for Achieving Alpha(一)
1. 怎么构建测试
所有的测试五等分,表明我们的回测的universe被分为五个组,根据我们要测试的公司因子的值。
Quintiles provide a clear answer to that question: if a strategy works, the top quintile should outperform, the bottom quintile should underperform, and there should be some linearity of returns among the qunintiles in between.
2. The backtest summary
(1) The years over which the test returns were calculated.
(2) Compound annual growth rates by quintile, based on the annually run portfolio returns.
(3) Average excess returns versus our Backtest Universe
(4) The percentage of one-year periods that the strategy outperforms the Universe.
(5) The percentage of rolling three-year periods that the stragety outperforms the Universe.
(6) The maximum gain realized over any one-year period
(7) The maximum loss sustained over any one-year period
(8) Sharpe ration of qunitile returns.
(9) The standard deviation of quintile returns.
(10) Beta of quintile returns versus the Backtest Universe. This represents a measure of how volatile a strategy is relative to the overall Universe. A number greater than 1 indicates a strategy that is more volatile than the Universe, while a number less than 1 indicates a strategy that is less volatile.
(11) Alpha of quintile returns versus Backtest Universe.
(12) Average portfolio size.
(13) Average number of companies outperforming.
(14) Average number of companies underperforming.
(15) The median portfolio value of the first factor used in this strategy.
(16) The median portfolio value of the second factor used in the strategy.
(17) The average market capitalization of the portfolios by quintile over the testing period.
3. Benchmarks
A quantitative strategy that works should have all or most of the following characteristics.
(1) Significant outperformance for the top quintile. For single-factor strategies, which have large average portfolio sizes, I like to see at least a 2% average excess return for the top quintile versus the Universe. For more focused, two-factor strategies, excess returns of 4% or more are preferable.
(2) Significant underperformance of the bottom qunitile.
(3) Good linearity of excess returns among the quintiles.
(4) Strong consistency of returns over time.
(5) Low volatility and low maximum loss for the top quintile/high volatility and high maximum loss for the bottom quintile. Both the Sharpe ration and Alpha can be used to provide an idea of a stragety's risk-adjusted returns, where risk is represented by volatility.