Frequent trend reversals
In the "Breaking Bad Trends" study, Research Affiliates examines various equity, bond, commodity and currency markets for characteristics of trend breaks. The total period under consideration ranges from 1971 to 2019, although the individual data series in some cases begin later in this period.
The central result of the investigations is that there have been more frequent trend reversals in the last ten years than before. This helps to explain the weak trend following returns of the last decade. In concrete terms, a turning point in the studies is defined as a month in which the short-term momentum of a market (1 month) has a different sign than the long-term momentum (12 months).
The following graph shows a negative correlation between the number of turning points (x-axis) and the risk-adjusted return of a 12-month trend-following strategy. For each market examined, the number of turning points (months with different signs for 1-month and 12-month returns) was calculated in each year. The strategy was assumed to be a classic, static trend following strategy: Long (short), if 12-month return is positive (negative). The calculations were equally weighted, both within and between asset classes. In addition, the portfolio returns were normalized to an annualized volatility of ten percent. Period: 1990-2019.
The number of turning points has a significant influence according to the studies: a multi-asset trend-following portfolio, normalized to an annualized volatility of 10 percent over the last 30 years, would lose about 9.2 percent of its annual return if the number of trend reversals was increased by one standard deviation.
A possible solution could be to use faster trend signals, i.e. to use shorter lookback periods than 12 months. However, the authors write that this is not a good idea. This would not solve the problem, but rather make it worse: This would increase the probability of bad trades, since faster signals contain an (even) higher proportion of noise.