Dear All,
Two tweaks are going to be released tonight.
- SHARES FIX
Back in August we released a new function SharesCur(bars) . It pulls in the number of shares outstanding from a daily feed and it’s more “fresh” than the number of shares from financial statements. For example it updates mid-quarter with information from different sources. We also use this function in the rare cases when a preliminary report comes in with number of shares missing before falling back to other types of shares or prior periods (see the documentation for the shares functions).
We noticed a problem though with companies that have multiple share classes. SharesCur will not include those ( it only includes the # of shares of the primary issue for a company). Before we can reliably use SharesCur as a replacement for outstanding shares from filing statements we need to make sure we include the other share classes. It’s a big project but once completed it will generate better , more up-to-date values for things like MarketCap, Pr2Sales, etc.
For the reasons above SharesCur will not be used for now as a fallback. It will still be available stand alone, but just keep in mind that it’s only pulling in one share class, and it will change in the future to represent the total most current # of shares outstanding
PS. we noticed these problem with Facebook. During a weekend of a preliminarey report back in 2012 the # of shares was N/A and we picked up the shares from SharesCur. However FB at that time had a lot of private shares, that are not traded, and SharesCur returned about 1/2 of the number of shares needed to calculate the MarketCap resulting in a cap of 20B vs 40B
- GAPS IN FINANCIAL STATEMENTS
When we create a point-in-time snapshot we build a list of 11 annuals and 25 interims that are “visible” for that as-of-date. In 99.9% of the cases this builds a list of contiguos statements. For example in the annuals you get 2012,2011,2010,2009, etc. However we were not checking for continuity. If a company stops reporting for a couple of years you could have 2012,2008,2007. This makes no sense for our purposes. So when we build this statement list we now stop if there’s no continuity.
Very sorry for the inconvenience. We doubt the above changes will materially change backtests.
Regards
