For a deeper look into our Eikon Data API, look into:

Overview |  Quickstart |  Documentation |  Downloads |  Tutorials |  Articles

question

Upvotes
Accepted
renee.yao avatar image
24 ●5 ●6 ●11

Too many NaNs in time series data

Hi there

I am using eikon scripting API to download minute-by-minute data (such as open, close) for about 2000 US stocks. I was able to download the data OK, with commands like below:

ek.get_timeseries(chunk, fields=field, start_date=today, end_date=dt.datetime.utcnow(), interval='minute', corax='adjusted')

here 'chunk' is a list of 1000 RICs that corresponding to the US stocks of interest. However, the return data contain a lot of NaNs, even for stocks of high liquidity (such as 'AAPL.O'). For example, if the start time is 9:30 of a trading day, and the end time is 16:00 of the same day, most of the data before 15:00 are NaN.

Am I doing something wrong, such as passing too many RICs (1000 in current) case?

Thanks!

eikoneikon-data-apiworkspaceworkspace-data-apirefinitiv-dataplatform-eikonpython
icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

2 Answers

Upvote
Accepted
pierre.faurel avatar image
4.3k ●2 ●4 ●5

Hi,

The TimeSeries response is limited to 3000 points, all values that are out of this limit are set to NaN.

If you want to stay under this limit (and avoid NaN values), you should decrease the RIC list and/or define a lower range of date.

icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

Upvote
renee.yao avatar image
24 ●5 ●6 ●11

Hi Pierre

Thanks for the answer @pierre.faurel, it didn't give me an error message when I'm asking more than 3000 points, hence I didn't realize the root cause.

As a workaround, I can reduce the number of RICs in one call, but I still like to have minute-by-minute resolution, for one day. Given the 6.5 hour trading day in US market, that's 390 points (max) per RIC. This lead to 7 RICs per ek.get_timeseries() call. So if I want to download all the ~2000 RICs that I am interested, I shall do ~300 calls in series? Shall I put a 'sleep' period between each call, or can I do those calls with multiprocess?

Thanks!

icon clock
10 |1500

Up to 2 attachments (including images) can be used with a maximum of 5.0 MiB each and 10.0 MiB total.

Click below to post an Idea Post Idea