Friday, 20 July 2012
Seasonals and calendar effects: Swinkels and van Vliet
1. Halloween effect (sell in May and go away but remember to come back in September (a bit adjusted): be invested only in the after Halloween period from November 1 to the witches sabbath of April 30 and
2. the end of the month effect TOM (at the last and first day of the month returns are higher).
All other calendar effects studied like several holidays, the January- effect, weekends etc were not strong enough to get really significant results.
When you try to measure those effects you must correct for a lot of double counting, many holidays are in the Halloween zone, also January and when you correct for Halloween+ end of month nothing significant remains.
Trading at the end/ first day of the month that is not something that will make administrators happy and they will prevent institutional investors to exploit TOM, especially around January 1. That could be a reason for the surprisingly strong results of TOM.
When you are in the market outside the Halloween and TOM periods it seems you cannot expect positive returns for equities when you extrapolates the results from 1963-2011. Even when you try to be clever by investing in value stocks or small caps that will not help to get positive results in those disaster periods (large caps are safer, but still not winning).
Warning: the studied effects are anomalies and these have often the bad property to vanish suddenly because they are arbitraged away. Not that long ago the January-effect was documented to be by far the strongest positive calendar effect and now it is only believed a bit for small stocks: the now profitable strategies are anomalies and they can suddenly stop to occur. Not that long ago the January effect was seen as the strongest calendar effect, especially for small caps (that small cap thing seems to be still all right).