Question on how much historical data is available. I know it says we can backtest starting 1/2/1999.
But if we rank stocks by 5 year earnings growth starting on 1/2/1999, will there be any data available to rank by? Or would we have to start in 2004 to look at last 5 years of earnings growth?
Similarly, if we want to rank by the last 10 years of sales growth, is the earliest we can start backtesting in 2009?
Thanks!
We have data going back to 1989 behind the scenes. Anything that looks backwards starting in 1999 is based on that additional data. So you can do both of your examples on 1/2/1999 and expect the results to be correct.
Perfect. Thank you!
Good info. Thanks for that. I’d been starting many of the backtests around 2005 in part to account for lag.
Is it possible that the 1989 to 1998 data could be “released” for backtesting, understanding that lagged data would not be limited or unavailable?
Hugh
Is this true for analyst estimate revisions or just for “hard” data contained in SEC reports and price and volume numbers?
atw: No, sorry. The design of the site is pretty much built around having a decade prior to the start behind the scenes.
o806: Properly speaking, the answer is going to depend on the exact data point.
For “hard” data, it is simply true. Pricing data and the full CompuStat fundamentals database go back to the 1950s, as I recall. (We don’t have the full database, just to head that off.) 
For the more peripheral data, it’s not necessarily the case more due to the history of data collection than actual history. Nobody thought to keep some of these things way back when, so we don’t have it today.
For specific examples, estimates revisions start in 2000, institutional trading starts in 2004, and IADs were retained starting about 2011.