In my Active Watch List, all of the holdings are noted as being held for more than 45,000 days.
I created a simulation that purchases stocks whose close(1) is 30 times greater than close(2) and sells when nodays>2. The annual return of 85% is astonishingly high. Of course I don't believe there could be that many instances - I had to cut the simulation that begins in 1999 short at 2018 because $100,000 had grown to more than $1 trillion if it runs longer - and I don't believe that buying stocks that have appreciated by 30x on the following day's close and holding for 2 days could even be profitable, much less incredibly so. I have created a public sim called "Data Test - Odd Results." https://www.portfolio123.com/port_summary.jsp?portid=1861250
Thank you very much for checking this out, and please advise. I am a huge fan of p123.
Most likely it has to do with the universe you are using in that simulation - that has no liquidity constraints. Pretty sure it is buying some illiquid nano stocks that in reality were unrealistic to buy.
Running it in an Easy to Trade universe for example, will lead to different results (assuming it runs at all).
Fun stuff though :D. I think you are aiming at post earnings announcement drift. If you look around on the forums, you can find some users that have experimented with that as well.
Just to be clear, I am definitely not trying to develop a trading strategy around this data. And it’s hard to think of a less efficient means of identifying possible data errors than using sims or even screens this way.
But these results are concerning, don’t you think? Particularly as my attempt is not comprehensive in the least.