I’ve started to write down some of the simple p123 codes of what are supposed to be studies that have yielded good returns: Open Source Cross-Sectional Asset Pricing - Google Sheets
One of them is the study of Heston and Sadka (2008) -1965-2002 with det description Average return in the same monthin the previous year.
But I just wanted to check if there might be something wrong with the code since it really didn’t give a very good return: 100*((Close(230)-Close(250)) / Close(250)) > 25 and MedianDailyTot(120)>( 70* 1000) and StaleStmt = 0
Here I have used 25% but have tried with different ones without it giving particularly good results.
Thinking about the question I believe that the assertion being made is different that of the initial study.
The initial study asserts that a stock which has a positive return in the reference period will have a positive return in the “current” period. A moderately strong assertion.
The proposed assertion is that a right tail positive return (outlier?) in the reference period will have a large positive return in the “current” period. And in looking for cumulative outperformance that it will repeat. A quite strong assertion.
Not having run any scenarios, I would be interested in testing eliminating reference period outliers to see if that improved results.
Cheers,
Rich
