Our team is building a credit-rating prediction model that uses focal
companies' supplier and customer financial data. Between
February and April 2026 we used the Refinitiv Workspace API
(refinitiv-data Python library) to pull:
- Phase 2: value-chain partner discovery via
get_data(universe=[...focal RIC...], fields='TR.SCRelationship*')
for 1,662 focal US-listed companies.
→ returned 13,265 unique supply-chain partners - Phase 3: quarterly financial pull for those partners via
get_data(universe=[...partner identifier...], fields='TR.F.*')
for the period 2000-Q1 to 2025-Q3 (61 quarterly financial fields).
Out of the 11,278 partners we attempted to pull financials for, the
outcomes were:
| outcome | count | meaning |
|---|---:|---|
| done — full quarterly financials returned | 5,269 (46.7%) | Usable, no help needed |
| no_data — RIC valid but Refinitiv returned blank | 931 (8.3%) | Need confirmation these truly have no data |
| deferred — OrgID only, no tradeable RIC available | 3,638 (32.3%) | Need RIC or alternative identifier |
| error — API returned an error during batched pull | 1,440 (12.8%) | Need root cause + retry guidance |
We would like the helpdesk's assistance recovering data for the 6,009 partners
in the bottom three buckets. Detail and asks below.
Bucket 1 — deferred (3,638 partners) — Need RICs or alternative identifiers
These companies were returned by Refinitiv's value-chain API as supply-chain
partners, but with only a numeric organisation ID (e.g. 5034750648) and
no tradeable Reuters Instrument Code (RIC). Our get_data calls against the
financial fields require a RIC.
Examples:
| OrgID | Company name |
|---|---|
| 5034750648 | Madrilena Red de Gas SAU |
| 5040225699 | Veeam Software Group GmbH |
| 4298209955 | Black & Veatch Holding Co |
Our questions:
- For each OrgID in the attached list deferred_partners.csv, can you
provide the corresponding RIC (or any other identifier such as ISIN,
SEDOL, or Refinitiv permID) that would allow TR.F.* financial pulls? - If a partner is genuinely a private company with no listed instrument
and no available financials, can you confirm that explicitly so we can
exclude them with certainty? - Is there a documented Refinitiv API method to convert OrgIDs to RICs
programmatically that we should be using? (We tried
symbology conversion via lseg.data.discovery.search but with mixed
results.)
Bucket 2 — error (1,440 partners) — Need root cause for API errors
Calls were batched as 20 partners × 31 financial fields × 16 quarters per
request. Most errors came back as:
Error code -1 | Unable to collect data for the field 'TR.F.TotRevenue'
and some specific identifier(s).
Requested universes: ['4296309931', '4296353850', ...20 IDs...]
Requested fields: ['TR.F.TOTREVENUE(PERIOD=FQ12016,FRQ=FQ,SCALE=6,CURN=USD)', ...]
We were not able to determine from the error which of the 20 identifiers
in the batch caused the failure, nor which of the 31 fields. Examples of
affected partners:
| Identifier | Company name |
|---|---|
| 4296309931 | Huawei Technologies Co Ltd |
| 4296353850 | Booz Allen Hamilton Inc |
| 5000144489 | Unison Industries LLC |
Our questions:
- Can the helpdesk identify, for each batch in the attached
error_partners.csv, which specific identifier(s) caused the
"Unable to collect data" error? - Is there a Refinitiv API setting / call pattern that would return
per-identifier errors rather than collapsing the batch? - For the affected partners individually, do you have data available
under a different identifier or field naming convention? - Several batches reference periods as early as FQ12000. Could the
error be that Refinitiv has no historical data going back that far
for these specific partners? If so, what is the earliest available
period per identifier
What we will do with the recovered data
Any RICs / identifiers you can provide for the deferred and error buckets
will be re-pulled through the same batched TR.F.* pipeline. This is for
internal credit-rating research only; no client-facing redistribution. We
estimate this will recover financial data for roughly 30-50% of the 6,009
missing partners, which would meaningfully improve the supply-chain
coverage of our rating prediction model.
If the helpdesk needs a sample of our successful pull pattern (the 5,269
done partners) to better diagnose what is different about the missing
ones, we are happy to share a sample script and field list.
Attached files
- deferred_partners.csv — 3,638 rows: lookup_id, company_name, identifier
- error_partners.csv — 1,440 rows: lookup_id, company_name, identifier, error_message_first_50_chars
- no_data_partners.csv — 931 rows: lookup_id, company_name, ric, identifier
Total: 6,009 partners requiring helpdesk attention.
Thanks very much in advance — happy to jump on a call if anything in this
note needs clarifying.
