I am relatively new to P123, so maybe it is something I am doing wrong, but when running the simulated strategies for Sentiment ranking I am getting different results now (2022 Aug 02/03) than I got back on 2022 Jul 10, but only for years 2014 and later. For years 2002 through 2013 I get the same result. I have checked every parameter I can think that I may have changed and cannot find anything different. I am running the Core:Sentiment ranking system, holding 5 stocks in my portfolio, and using the S&P 1500 as my universe. (Since I am still learning and exploring I wanted to keep things simple and add complexity as I go.) I have simulated results for rolling 3-year periods for 2002 Jan 01 … 2005 Jan 01, 2003 Jan 01 … 2006 Jan 01, etc. As I said, all of the 3 year rolling periods are the same until the year 2014, and then they are different. I have attached a text file identifying all of the years that are different, along with sample outputs for the 3-year period of 2013…2015.
I don’t think you’re doing anything wrong. In the past four weeks some past estimate data might have changed slightly, causing you to buy or sell one or two different stocks. (Once in a while FactSet will notice a mistake they made with some estimate data and correct it, or remove or add an analyst. With FactSet, historical data is not completely static.)