KOF and data integrity for ADRs

major discrepancy in market cap and Shares outstanding

P123 - 5.27B, 53M shares
Yahoo finance - 20.9B; 210.7M

Which is right? This happened recently on another stock which was an ADR. Makes me wonder about rankings based on this data.

It shows up really cheap in core combination ranking. When I use the actual data to do a dcf - I get a per share value around 60, as opposed to 240 with the data currently showing.

Koyfin shows KOF shares outstanding as 210.1M

Can someone from P123 look into this please? @danparquette, @yuvaltaylor, @marco ?

Investigating. Thanks for reporting it

Any update here? thanks

Asking FactSet.

They give use SharesQ=52.52 and SharesFDQ=210.08. The company has two other share classes and SharesQ should include those. Compustat has both set to 210.08. The FactSet workstation shows a cap 20.9B since it includes all share classes, but in our datafeed there is a cap of 5.2B.

So at the moment it appears to be a data error.

Thanks Marco. That brings up another point. can we also show fully diluted share count in the snapshot page? Ideally stock based comp in panels too.

Is this an issue for ADRs in general, or just this one?

I don’t know, i suspect it could also be due to different share classes also. I am double checking every ranking output. I am mostly discretionary so I can do that. If someone is running an automated portfolio, it can be a cause for concern.

1 Like

Will do for SharesFD. We already have a task to add a few other factors to the Snapshot.

We do not have a “stock based comp” at the moment.


1 Like

FYI. We’re still working with FactSet re. this. Other stocks are problematic too, maybe around 5%, where the numbers of shares used should come from a different field in the data that we get. But it’s not at all clear what logic should be used. It seems that their own product, the FactSet Workstation, uses additional data that we don’t get to calculate market cap.

But like I said, while understandably frustrating, it is a relatively small number of stocks.

It may be a small number, but it affects a couple of my portfolios targeting tiny stocks.

I appreciate that you’re looking into it.

Partial good news.

FactSet agreed to deliver to us a new table with precise historical market cap values, which is quite hard to calculate precisely in many cases. This will also fix stocks that are impossible for us to compute like IBKR (they have these weird non traded Holdings Membership Interests shares)

But the new data only starts from 12/31/2015.

Regardless we will use it and fallback to our best calculation prior to that. We're also evaluating how best to use this data for other ratios like valuation ratios. We will deploy it soon, before AI, since it will cause some differences in the past.

Let us know if u have questions.

Thanks Marco!

You mentioned them correcting historical market cap data. What about current market cap data going forward, can they correct it? There are often issues with market cap data for nano-caps.