I re-downloaded some data through the API and compared it with the exact same data pull from May of this year, and all of the factors which include current assets or current liabilities or working capital are materially different.
Can you give us some specific examples? Are you sure you're examining data from the exact same quarter? WorkCapQ, for instance, will of course be different now than in May.
I pulled down 10 years (2010 - 2019) of data for all my factors in May. I pulled down the same 10 years (2010 - 2019) of data for all my factors this week. I calculated the correlation between all 10 years of data from May vs. Aug for each factor individually. For every other factor, outside of those involving current assets / liabilities, the correlation was either 1.0 or 0.9999 (ie. the data is precisely the same). Correlation for factors involving current assets and liabilities was much lower at 0.75.
Those items are N/A for all financial companies. Might that have something to do with it?
No, why would that affect a data pull from May differently than August? I downloaded the same data using the exact same parameters, and the data was different. The historical numbers in your database have changed.
There are certain examples where it was NAN before (ranking of 50) and now has a rank of either 0 or 100.... for example, when I pulled the data on May 28, JBGS curratioq for 2025-05-05 was 49.99 and is now 0.37 when I pull the data for that same date.
OK so it seems the number of NAN's has changed significantly. For the exact same universe, when I pulled the data on May 28, there were 562 NaNs for curratioq, and now when I pull the data there are 68. Because the number of NaNs changes, the ranking for all the non-NaNs changes as well, although order is preserved (I think).
EDIT: Once I remove NaNs (aka 49.99) from the old data, and new data, the spearman correlation is 1 so order is preserved - it is precisely an issue of the NaN's.
You just need to explain how / why the number of NaN's changed so significantly for curratioq since end of May. Maybe its on FactSet's side?
Preliminary findings...
Problem with JBGS seems to be the "Assets held for sale", which is considered a current asset, and it is listed as "---" in their March filing . I get very suspicious when I see "---" in a filing. Sometimes vendors convert "---" to a 0 sometimes it's an N/A. Either way, AstCurQ should be around 314,795 for March but we are getting a 0.
Their next filing "Assets held for sale" is 151,082, and AstCurQ is around 439,125.
So yes, it smells like some data changed, and that we need to report some data errors.
You have other examples?
Thanks
March 10Q
JBG SMITH Properties (Form: 10-Q, Received: 04/29/2025 16:17:21)
June 10Q
JBG SMITH Properties (Form: 10-Q, Received: 07/29/2025 16:18:41)
Sure, on the same date (2025-05-05), the following tickers went from percentile 50 (ie. NaN) to around 0:
JBGS 49.99 to 0.37
AI:CDN 49.99 to 0.39
GEO 49.99 to 0.52
NNN 49.99 to 0.72
EQR 49.99 to 1.06
CIM 49.99 to 1.30
IVR 49.99 to 1.72
Also Yuval was correct, they are mostly sector 30. So Factset must have changed how they calculate curr assets / curr liabilities for Financials... I'm surprised they don't send out a notification before they make a change like this?
It appears to me that you switched from NAs neutral to NAs negative.
Both pulls were done with NA set to neutral:
RankData = client.rank_ranks({"pitMethod": "Prelim",'rankingSystem': 'all v9', 'asOfDt': RunDate, 'universe': 'all na',"precision":4,"nodeDetails": 'factor','rankingMethod':4})
rankingMethod: 4 should be neutral?
Also I'm getting 100s and 0s, not just 0s (i just gave 0s as the example). Its because there are now lots of examples with either current assets = 0 or current liabilities = 0. So for curratioq you either get 0 or infinity for those.
Now I see it. CurRatioQ, LiabCurQ, and AstCurQ are all now populated for REITs and banks, which is quite unexpected. On FactSet's online assistant site they detail how they calculate current assets and liabilities for banks. If you're interested, I can post the details.
1 Like
That'd be great, and also the exact industry codes that were affected.
This is what they have on their website. They don't mention industry codes.
Online Assistant _ FF_ASSETS_CURR.pdf (124.1 KB)
Online Assistant _ FF_LIABS_CURR.pdf (122.7 KB)
Perfect thanks for your help guys, much appreciated.
Had to argue with Factset a bunch re. JBGS Current Assets being 0 for 3/31/2025.
Their initial response was kind of hilarious: they included a screenshot of their documentation, and doubled down with
FF_ASSETS_CURR represents all assets that are expected to be used up or sold in the current period. In the company's quarterly report, this is described as Assets held for sale. Upon reviewing the 10-Q source you shared, the company has not reported assets held for sale, which is why this is showing as 0 for the 3/31/2025 period.
They highlighted the "----" entry for "Assets held for sale." and said that 0 is the correct answer.
So I had to show them that their methodology for other quarters includes other things for current assets, like "Cash" (duh!). I sent them the image below with what they have for 6/30/2025 and what 3/31/2025 should look like.
Seems like I was right. The "---" is being interpreted as N/A and their logic simply gives up, resolves to 0, and skips all the other current assets.
It was finally escalated to someone that understands a bit more and it's being fixed . Not sure what this explanation means, and if it means they will fix their faulty logic 
The collections are reflecting fine for FF_ASSETS_CURR for 3/2025. Due to technical issues data is not reflecting incorrect in the product. The team has re-exported the collected data to fix the issue and we will update you once we receive new information from the team.
Interesting, it seems like they've switched from manually ingesting the data with humans to an NLP. Or have they always been using NLP?
Should be easy for them to fix I guess? For now, I'm just overridingany current assets / current liabilities factors to NaN for banking and REITs.
My favorite is “due to technical issues data is not reflecting incorrect”
1 Like