Portfolio construction is just as much art as it is science. Part of the art comes from experience: intuitively understanding the implications design decisions can have on portfolio performance due to model error.
For example, in the past we've seen portfolios that employ trend models pivot from equity beta exposure to long-duration treasuries. This switch will be rewarded in trending markets when the negative correlation will not only lead to relative outperformance from removing equity beta, but excess return from the positive return from treasuries. In mean-reversionary markets, however, whipsaw cuts both ways. Q4 2011 was particularly difficult on these sorts of strategies.
Enter BTAL, QuantShare's Market Neutral Anti-Beta fund. The strategy seeks to capture the spread between high beta and low beta securities, going long low-beta and short high-beta, resulting in a consistently negative beta instrument.
An interesting aspect of BTAL is that it tends to have an asymmetric up-capture / down-capture ratio relative to equity markets. In other words, BTAL will gain more when the market goes down than it will lose when the market goes up. This makes it an interesting hedging vehicle because we're taking an asymmetric bet (our reward is greater than our risk) and therefore the whipsaw trades will be less damaging.
Consider a constantly hedged equity portfolio: 50% S&P 500 (SPY) and 50% BTAL rebalanced weekly. Over the long run, this portfolio should give us 0.25 equity beta exposure (estimating that the long-run beta of BTAL is -0.5). Why have a 50/50 portfolio instead of just 25% exposure to SPY? If our expectation is that BTAL's beta will be less negative during calm markets and more negative during turbulent markets (theory: people will sell low beta securities and purchase high beta securities in calm markets to replicate leverage; people will flee high beta securities and buy low volatility securities in turbulent markets; the selling will drive up volatility in that leg of the portfolio), then we will get extra beta exposure when we want it and reduced exposure when we don't.
The issue is that the asymmetric up/down capture of BTAL is not constant. We may, therefore, put on a hedge using BTAL during a whipsaw environment when our risk is greater than our reward. However, it does seem to be fairly persistent (see the 63-day up-capture / down-capture spread ratio in the above graph) -- so what we can do is measure the spread between up-capture and down-capture and only apply the hedge when it is directionally in our favor. When it isn't, we can simply reduce our position from 50% to 25% in SPY.
We find that selectively applying BTAL as a hedge -- removing it when the asymmetric hedge is not in our favor -- dramatically increases total returns. This is most evident in 2009, when BTAL suffered dramatic drawdowns due to a significant beta unwind, but because the unwind was so violent the asymmetry flipped and it was removed from the portfolio. This is the "art" of portfolio development: creating quantitative rules that capture an intuitive factor and behaves how we would expect it to.