Using daily returns that have been winsorized relative to the entire universeU 17 ,
Monte Carlo portfolios for the two strategies are recorded in Figure 9.5. The good news
is that every possible sentiment reversal portfolio outperforms the S&P 500; the bad
news is that the excess returns have dropped a significant amount. The best outcome
occurs at the 5th starting point with no entry interval, where the annualized return is
5.73%, yielding an excess return of 11.38% (still not quite as glamorous as the 16%
excess return obtained by an equally weighted average over the entire universe). If the
annualized returns are averaged over every starting point, the strategy with no entry
interval gives 1.67% while the 4-month entry interval strategy gives 0.80%; or excess
returns of 6.46% and 7.32%, respectively. The difference in performance of the two
strategies might be expected, given the timeframe over which performance is computed.
The strategy with more cash will likely outperform its rival in such bearish market
environments.
The detraction in excess returns from 16% in the equally weighted case to 7% in the
portfolio-simulated case is understandable. It speaks to the distribution of anomalistic
opportunities and realistic expected returns. On the other hand, it might be said that this
is a commentary on the market-specific characteristics of sentiment reversals. In bear
markets, sentiment reversals still outperform the market; however, they do even better in
bull markets (clearly, the starting points used in the 2000–2008 Monte Carlo simulations
were good stocks to own in the late 1990s in order to obtain an excess return of 16% over
the entire universe). One might hypothesize that the information lag supposedly driving
the anomaly is less pronounced in bear markets. That is to say, perhaps investors are242 News and abnormal returnsFigure 9.5.Nine-year annualized returns over 2000–2008 for 10-stock Monte Carlo portfolios of
U 17 sentiment reversals as a function of starting points.