IMPORTANT: fixes that affect Surprises data (may affect past results)

Announcements
1

Dear All,

We have released a new version to address some lingering problems, and some new ones due to a recent change in Compustat. It affects the surprise functions and factors in this section of the reference: ESTIMATES->SURPRISE & ACTUALS.

There are 4 different fixes:

  1. Estimates are on the CapitalIQ side which, sometimes, does different things than the Compustat side for major corporate actiosn. For example when FedEx & RPS became one company in 1998 Compustat did not start a new entity, CapitalIQ did. We were not handling cases like this. The end result is that for Fedex we ended up having N/A’s for the surprises because we were picking up the defunct id. It’s a pretty rare case, about 100 or so companies, of which FDX is the main one (AET is another).

It is now corrected and you can verify it in the DATA->SNAPSHOT page. Overall the impact when you re-run a sim should not be meaningful, some small random noise, since the companies affected include ones that have surprised and some that have not. I re-ran the Chaikin Indicators model of P123 , which uses Surprises in the ranking system, and the differences was the fix did 0.2% annualized better.

  1. Since we have USA & Canada, and we have companies listed in both, there was an assumption that the consensus estimates were one. They are not, and sometimes very different when, for example, one analysts covers in the US & 15 in CN. Sometimes there are consensus estimates attached to the USA shares. Sometimes they are attached to the CAN shares. And sometimes there are estimates in both, with varying number of analysts making up the consensus. We are now using the consensus that has the highest number of analysts, and USA when there’s a tie. Naturally after appropriate currency conversion. The impact of this should be minimal.

NOTE: not sure why there’s isn’t a data provider that creates a “global” estimate since there’s only one company. For example according to CapitalIq RYANAir met analysts expectations in Europe, but US analysts it disappointed 18%. It’s technically right , but wouldn’t it be more right with one global consensus surprise metric?

  1. ADRs (1). Our database has estimates from all over the world that, if not USA based, do not account for the ADR ratio. We did not know this. This is now corrected. The effect should be minimal in the context of surprise %, and since it only affects ADRs

  2. ADRs (2). ADRs had the same problem as Canada (2) . We applied the same logic.

Very sorry for the inconvenience. Let us know of issues & concerns.

Thank You

2

Thanks Marco. That was quick!

3

No , not really. It was a complicated , ongoing issue.

4

Quick since Iavantis post. In any case thank you.

5

Iavanti’s is a different problem :frowning: