Read on Twitter
1/ Trend following just once a month will suffer a lot from timing luck. The popular end-of-month check is sometimes the best, sometimes the worst of 20 possible alternatives.
2/ Comparing different schedules for different trend following lookback periods (21-day out to 378-day) shows that longer lookback periods suffer less from timing luck (lower range of possible returns), but timing luck still has a big effect.
3/ Backtests should use blended portfolios to eliminate this role of luck. This shows the best trend-following lookback period would have been around 13 months.
4/ Blending different timing schedules offers some tax benefit, since particular shares can be sold based on optimal holding period or gains/losses.
Within a blended portfolio, the different timing schedules will often show large differences in holding period.
Within a blended portfolio, the different timing schedules will often show large differences in holding period.
5/ A blended portfolio also makes a quarterly trend portfolio more viable. Some of the portfolio will catch changes in trend, but whipsaws will be less painful.
The blended quarterly schedule suffers little loss in return vs. monthly while turning over a lot less.
The blended quarterly schedule suffers little loss in return vs. monthly while turning over a lot less.
6/ I look at these benefits of blending overlapping portfolios here: https://justinczyszczewski.com/trend-and-blend/
