1/ Implementing Momentum: What Have We Learned? (Ross, Moskowitz, Israel, Serban)
"Live momentum portfolios are capable of capturing the momentum premium, even after expenses, trading costs, taxes, & other frictions associated with real-life portfolios."
2/ "We decompose the performance differences between the live momentum strategies and the theoretical AQR Momentum Indices.
"We expect that trading costs and expenses will always be a drag on performance, but ‘smarter’ portfolio construction decisions should offset this."
3/ "Momentum's historical excess return is 1.0-1.5%/yr. Even if transaction costs were twice as high as realized, they may only put a small dent in performance.
"These results contrast with papers that estimates costs based on liquidity-demanding orders for immediate execution."
4/ "While the momentum style has had a tough period over the past seven years, transaction costs were not the culprit.
"Despite the higher turnover relative to other style strategies, such as value and defensive, momentum still survives transaction costs."
5/ "Tax-managed strategies have captured a similar return premium as the regular MOM strategies, further evidenced by the high correlation of excess returns and low realized tracking error.
"They also have realized similar or lower taxes than the corresponding passive indices."
6/ "We expect and find that the costs of implementation are small, while the benefits of portfolio construction are significant. A live strategy captures the theoretical momentum premium without incurring substantial real-world costs and potentially adds value."
One way to be a liquidity provider is to be an endowment: no leverage and infinite time to wait for convergence.
Another is to be simultaneously diversified/hedged, such that divergence has a payoff and allows you to hold on to convergence trades with very high expected returns.
2/ This resembles combining multiple measures of ST reversal, momentum, and LT reversal (forecasts determined by walking forward rather than using signs from the full sample).
Unlike normal moving average signals, these are *cross-sectional.* More below:
3/ Unsurprisingly, the Trend factor formed by this approach outperforms benchmarks in terms of both Sharpe ratio and tail metrics. It's combining momentum with two factors that are negatively correlated to it AND using multiple specifications.
2/ * July 1963 to June 2018
* Factors 1-6 are the most popular academic factors, followed by factors 7-14, then the 33 factors in the "other" group
* Long/short Fama-French portfolio construction (this leads to factors being dollar-neutral rather than beta-neutral)
3/ "When the market does great (months >1σ above the mean), most factors do not; when the market moves sideways, most factors deliver positive returns; and when the market falls >1σ, most factors are at their best. Similarly, on average, factors have higher premia in recessions."