You can cut down the number of requests by retrieving multiple fields in a single request. Using the following code I retrieve 13 years of daily price history for the constituents of S&P 500 in about 8 minutes. In this code spx is the list of RICs for the constituents of S&P 500.

ts = pd.DataFrame()
chunk_size = 50
for i in range(0, len(spx), chunk_size):
    rics = spx[i : i + chunk_size]
    df, err = ek.get_data(rics,
                      ['TR.CLOSEPRICE.date','TR.CLOSEPRICE',
                       'TR.HIGHPRICE','TR.LOWPRICE',
                       'TR.OPENPRICE','TR.ACCUMULATEDVOLUME'],
                     {'SDate':'-13Y', 'EDate':'0D'})
    ts = ts.append(df, sort=False)
ts

The resulting dataframe has 1.65 million rows. I must admit I have no first hand experience with Bloomberg. I can easily believe that using Bloomberg API one can retrieve all these timeseries in a single request. Then again adding the above code to break the list of instruments into chunks and to retrieve chunks in a loop is not a huge task, and you only need to do this once. As for the data retrieval time I find it very hard to believe that all this data can be retrieved in 5 seconds from any system. If it is not an exaggeration, that is mightily impressive. 8 minutes is not the best time you can possibly get out of Eikon, although further improvement would not be as straightforward as in the above code sample. And even with the max optimization I don't think we can get the retrieval time to significantly under a minute.

For the first problem, it could be limitations in Eikon Data APIs mentioned in the EIKON DATA API USAGE AND LIMITS GUIDELINE.

get_timeseries: The current limit value (10-Oct-2019) is 3,000 data points (rows) for interday intervals and 50,000 data points for intraday intervals. This limit applies to the whole request, whatever the number of requested instrument. 

For the second problem, the TR.PriceClose field is not available for .IBLEM0002. You can use the Data Item Browser (DIB) tool to verify it.

Thank you @irapongse.phuriphanvichai - a bit disappointing. This is much much less than what Bloomberg allows - making the API less useful for even a simple analysis such as the S&P500 10 years data: 500 x 10 x 250 = 1,250,000 data points or 417 requests - essentially looping over each constituent. Even just analyzing a single stock over the past 20 years needs 2 requests and a merge.

Question: How is the community dealing with this problem? Thank you very much in advance.

@HeikoRR
To retrieve timeseries of daily price history for exchange traded instruments including stocks and indices you can use get_data method with TR.CLOSEPRICE field.

Hi Alex, thank you for your reply.

get_timeseries():

I have answered you with respect to get_timeseries()

https://community.developers.refinitiv.com/questions/14078/data-limis-for-eikon-api-proxy.html?childToView=57588#comment-57588

There, the result for a OHLC, Volume, daily, 13 years for S&P 500 constituents were:

EIKON: 5000 requests and 2500 merging of time series, 27.8 minutes

BLOOMBERG: 1 request, no merging, 5 seconds

get_data():

I tried to do the same using get_data(). The situation is a bit better but still way worse than using the Bloomberg API: Assume again I want to do a quick analysis on the S&P500 with

Open Low High Close and Volume (5 timeseries), daily, starting in 2007 (13 years ago)

For the time period (2007-01-03 to 2020-03-25, 3332 days) I can only request 63 to 64 tickers in one request (I tried many combinations). That amounts to 500 * 5 / 63 = 40 requests. If I request more, get_data() returns NULL without an error message.

EIKON: 40 requests, no merging of time series

BLOOMBERG: 1 request, no merging, 5 seconds

Note: Some tickers such as .IBLEM0002 do not have the TR.PriceClose field and have to be retrieved using get_timeseries()

Please let me know if I miss anything?

Hi Alex,

Yes, thank you again. Sure it is easy to loop over chucks of rics.

But why having these restrictions in the first place? It would be helpful if you could internally see if these restrictions could be adapted to allow to retrieve reasonable amounts of data per day in a single request - especially the limits on get_timeseries are way to low.

Thank you in advance, Heiko